Australian Institute of Criminology

Skip to content

Prosecutorial decisions in adult sexual assault cases

Trends & issues in crime and criminal justice no. 291

Denise Lievore
ISBN 0 642 53865 4 ; ISSN 0817-8542
Canberra: Australian Institute of Criminology, January 2005

Foreword | Prosecution agencies are often criticised for their performance in prosecuting sexual assault. A lack of external transparency means there is little knowledge about the specific criteria used in decisions to proceed with or discontinue prosecutions. Understanding the factors that impact on the exercise of prosecutorial discretion therefore constitutes an important step towards improving criminal justice outcomes in sexual assault prosecutions. This paper analyses prosecutorial decisions to proceed with or discontinue prosecution in a sample of adult sexual assault cases. The results indicate that case decisions are primarily based on evidentiary considerations related to the ability to secure a conviction, but they also raise questions about the handling of cases involving prior relationships.

Toni Makkai
Acting Director

Prosecutors employed by the state and territory offices of the Director of Public Prosecutions (DPP) are vested with a range of discretionary powers in relation to prosecuting indictable offences. Some of the most important decisions include whether to proceed with a prosecution and whether to discontinue prosecution by either deciding not to find a bill of indictment (no bill), or declining to present an indictment to the court (nolle prosequi). Each DPP has developed a prosecution policy to guide prosecutors in making decisions about the conduct and disposition of cases. While there are differences in the various guidelines, they all identify two essential elements to be considered in determining whether to prosecute (Refshauge 2002). First, there must be sufficient evidence to justify prosecution and provide reasonable prospects of conviction. This assessment takes into account factors such as the competence, credibility and availability of witnesses, the admissibility of evidence, and any other factors that could affect the likelihood of a conviction. Once it is established that there is sufficient evidence, the second and overriding concern is whether it is in the public interest for the matter to proceed. The public interest is not a question of political or popular pressure; relevant factors include the seriousness and prevalence of the offence, factors related to the victim and the defendant,1 and the need to maintain public confidence in institutions such as the courts.

Few Australian studies have examined factors that predict prosecutorial decisions in adult sexual assault cases. A Queensland study (Briody 2002) and United States research show that case decisions are shaped by a complex interplay of structural and attitudinal factors (Kerstetter 1990; Kingsnorth, MacIntosh & Wentworth 1999; Spohn, Beichner & Davis-Frenzel 2002). Decisions are largely driven by legal considerations relating to the prospects of conviction, such as the severity of the crime, the type and strength of the evidence, and the defendant's culpability. The data also show that prosecutorial decisions are open to influence by variables that are extraneous to the legal elements of the case, including sociodemographic characteristics of victims and defendants and the victim-defendant relationship. These decisions often involve legal issues that are matters of professional judgment and require a degree of subjectivity, so prosecutors can have different views on a matter and some decisions may reflect personal biases.

An influential perspective on legal decision-making, derived from a study by Albonetti (1987), suggests that the relative weight given to legal and extralegal factors will vary, as prosecution decision-making is principally oriented towards eliminating uncertainty in pursuing cases that are likely to lead to a conviction. Prosecutors know that they cannot predict or control the behaviour of the defendant, defence counsel and jury, so the probability of proceeding increases in the presence of legal and extralegal factors that boost the likelihood of success and decrease uncertainty over potential outcomes (Albonetti 1987). Some studies have found that legally relevant factors are predictive of decisions in more serious sexual assaults, where prosecutors may have less scope to exercise discretion or to consider extralegal factors. Extralegal variables may be more influential when the charges are less serious or where the prospects of conviction are diminished, such as where there is a prior relationship between the victim and defendant (Spohn & Holleran 2001).

Various studies have found that 'stranger rapes' are more likely to proceed to trial and conviction than cases involving acquaintances or intimate partners (see for example, Harris & Grace 1999; Kingsnorth, MacIntosh & Wentworth 1999). This has given rise to concerns that legal judgments about the prospects of conviction may be filtered through stereotypical images of 'real rapes' and assumptions about the nature of heterosexual relationships (Stanko 1982). At the same time, the prosecution faces distinctly different issues when the victim and defendant are strangers as opposed to when they are acquainted. In the case of strangers, the assailant's identity is a central issue; but as the majority of victims and defendants have preexisting social or sexual relationships, the main issue of most trials is whether the sexual activity was consensual, or whether the defendant believed the victim was consenting (Bryden & Lengnick 1997). Researchers often conclude that attrition of sexual assault cases at the prosecution stage is primarily related to evidentiary matters, which are most difficult to resolve in the latter cases, as they usually come down to word against word with little or no corroborating evidence (Harris & Grace 1999). However, quantitative analyses cannot capture gradations of human interactions that impact on official decision-making, such as interactions between prosecutors and victims (Kerstetter 1990).

For example, the victim's willingness to prosecute is an important but not necessarily determinative factor in official decisions to invoke the criminal law. While the victim's willingness to proceed to trial is often regarded as 'an unproblematic, extrasystemic factor' (Kerstetter & van Winkle 1990: 269), it is more complex than a simple matter of volition. Research has demonstrated that victims' choices are influenced and sometimes manipulated by prosecutors (Kerstetter & van Winkle 1990). When there is a likelihood that a case will be discontinued, the prosecutor must establish a legal rationale for this decision. Rather than telling the victim what to do, the prosecutor may shape the victim's choices by reducing her options (Frohmann 1998).

This paper reports on selected findings from an Australian study analysing prosecutorial decisions made by DPP prosecutors in five jurisdictions: the Australian Capital Territory, New South Wales, the Northern Territory, Western Australia and Tasmania (Lievore 2004). The study focused on cases involving indictable sexual offences against adults. While most defendants were charged with other types of offences, including acts of indecency, assault and threats to kill, the analysis focuses on outcomes for sexual assault charges only. Statistical data relating to victim, defendant and case characteristics, and case outcomes were collected through a survey of 141 DPP case files, which related to 148 victims and 152 defendants. Six cases involved multiple victims and seven cases involved multiple defendants. No cases involved multiple victims and multiple defendants. Table 1 outlines the inclusion criteria for cases, relevant state/territory legislation and the variables recorded for analysis. Victim-based data are presented in Table 2 and defendant based data are outlined in Table 3. Due to inconsistent recording practices within and across DPPs, there are variations in the number of victims and defendants for whom information was recorded and not all variables were used in subsequent analyses.

Table 1: Inclusion criteria and variables recorded for case survey
Inclusion criteria
  • Cases were referred to the DPP between 1 July 1999 and 30 June 2001; they had since been finalised, either through a court determination or by discontinuance
  • The victims were adults (over the age of consent) and deemed competent to consent to sexual intercourse
  • The defendants were or would have been tried in the adult courts
  • When referred to the DPP, the primary charge was rape or an equivalent sexual assualt, corresponding to the Australian standard offence classification (ASOC) subdivision 0311 - aggravated sexual assault (ABS 1997: 34)
State/territory legislation
  • ACT Crimes Act 1900 ss 51-54 (sexual assault in the first degree, sexual assault in the second degree, sexual assault in the third degree, sexual intercourse without consent)
  • NSW Crimes Act 1900 ss 61I-61JA (sexual assault, aggravated sexual assault, aggravated sexual assault in company
  • NT Criminal Code Act s 192 (sexual intercourse and gross indecency without consent)
  • Tas. Criminal Code Act 1924 s 127A and s 185 (aggravated sexual assault and rape)
  • WA Criminal Code ss 325-328 (sexual penetration without consent, aggravated sexual penetration without consent, sexual coercion, aggravated sexual coercion)
Victim characteristics
  • Gender, age at time of offence, relationship status at time of offence, race/ethnicity, employment at time of offence, substance use at time of offence, to whom offence was reported
Defendant characteristics
  • Gender, age at time of offence, relationship status at time of offence, race/ethnicity, employment at time of offence, substance use at time of offence, criminal history, relationship to victim
Case characteristics
  • Number of victims, number of defendants, victim injury, additional evidence, weapon use, force, threat, victim nonconsent (physical or verbal), victim incapacitated, time taken to report offence
Case outcomes
  • Was case withdrawn?
  • If case withdrawn, at what stage?
  • Who initiated withdrawal?
  • Were charges negotiated?
  • Outcome for major sexual assault charge
Table 2 : Victim-based sociodemographic and case characteristics
Victim sociodemographic characteristicsn%
Source: Australian Institute of Criminology, Sexual assault DPP study, 1 July 1999 to 30 June 2001 [computer file]
Gender 148 100
Female 146 99
Male 2 1
Age group 140 100
16-24 56 40
25-44 63 45
45+ 21 15
Relationship status 122 100
Single 39 32
Partner 60 49
Separated, divorced, widowed 23 19
Race/ethnicity 71 100
Caucasian 20 28
Other* 51 72
Employment status 127 100
Employed 46 36
Not employed 81 64
Substance use 126 100
No 42 33
Yes 84 67
Victim-related case characteristicsn%
Source: Australian Institute of Criminology, Sexual assault DPP study, 1 July 1999 to 30 June 2001 [computer file]
Time taken to report assault 142 100
Within three hours 82 58
More than three hours and less than one week 40 28
More than one week 20 14
To whom assault was reported 141 100
Family/friend 62 44
Police/other authority 40 28
Other 39 27
Injury 148 100
No 72 48
Yes 76 52
Expressed non-consent 148 100
No 60 39
Yes 88 61
Incapacitated (unable to consent) 148 100
No 130 88
Yes 18 12
Table 3 : Defendant-based sociodemographic and case characteristics
Defendant sociodemographic characteristicsn%
* These figures reflect the high proportion of NT and WA cases in the sample
Source: Australian Institute of Criminology, Sexual assault DPP study, 1 July 1999 to 30 June 2001 [computer file]
Gender 152 100
Female 0 0
Male 152 100
Age group 151 100
16-24 49 32
25-44 89 58
45+ 13 9
Relationship status 28 00
Single 36 28
Partner 59 46
Separated, divorced, widowed 33 26
Race/ethnicity 134 100
Caucasian 54 40
Other* 80 60
Employment status 143 100
Employed 63 44
Not employed 80 56
Substance use 126 100
No 15 12
Yes 111 88
Defendant-related case characteristicsn %
Source: Australian Institute of Criminology, Sexual assault DPP study, 1 July 1999 to 30 June 2001 [computer file]
Criminal history 126 100
No priors 17 14
Sex and/or violence (includes 'other') 69 54
'Other' only (for example, drugs, property) 40 32
Relationship to victim 149 100
Stranger 35 24
Current partner 23 15
Former partner 17 11
Family 22 15
Other known 52 35
Weapon use 146 100
No 117 80
Yes 29 20
Physical force 146 100
No 41 28
Yes 105 72
Threat 146 100
No 102 70
Yes 42 30
Additional evidence 146 100
No 87 60
Yes 59 40

The sample comprised 17 cases from the ACT, 34 cases from NSW, 52 cases from the NT, 11 cases from Tasmania and 27 cases from WA. This represents all relevant cases identified for the reference period in the ACT, the NT and Tasmania. It was not possible to extract and analyse all relevant cases for NSW and WA. In these states, DPP staff selected up to 50 of the most recent cases that met the inclusion criteria. Some of these were subsequently excluded, as they did not meet the criteria.

Given the small numbers of cases for some jurisdictions, statistical analyses were conducted for the pooled sample only. This results in more weight being given to some jurisdictions than would be the case in a random sample. In particular, the NT and WA cases comprised 56 per cent of the sample and had significantly higher percentages of Indigenous defendants (amounting to 48 per cent of the total). This is expected, given the larger Indigenous populations in these jurisdictions, but the findings may be biased by the particular characteristics of these populations.

Preliminary analyses of cases involving differing numbers of victims or defendants showed that these decisions applied to the entire case, rather than individuals. Therefore, the following analyses were conducted on a case basis, by selecting the first victim and first defendant in each case and excluding data on additional individuals. As the majority of cases involved only one victim or defendant, this did not result in a substantial loss of information. Key findings are:

  • 62 per cent of cases in the sample either proceeded to trial or were finalised by way of a guilty plea;
  • 29 per cent of all cases proceeded to trial;
  • 38 per cent of these resulted in a guilty verdict (this amounts to 11 per cent of the total sample);
  • 33 per cent of cases were finalised by way of a guilty plea; and
  • 51 per cent of these followed negotiations to reduce the number or level of charges.

The data were analysed to determine which variables were associated with cases either proceeding or being discontinued, as well as the stage at which cases were withdrawn. To control for procedural differences between jurisdictions, cases that did not proceed were categorised as being withdrawn either prior to or after an indictment was filed. The basis and stage of case withdrawal is shown in Table 4. The 88 cases that proceeded to trial or sentencing represented 94 victims and 95 defendants; outcomes were available for 92 defendants. Bivariate analyses showed that there were statistically significant differences between cases that were withdrawn and cases that proceeded with respect to seven variables (Table 5). Most of these variables can be classified as legally relevant. For example, nonconsent - expressed either verbally or through physical resistance - is the basic substantive element of sexual assault. The victim's claim that she did not consent to sex is corroborated by injury, use of force, the severity of the assault and the availability of additional evidence linking the defendant to the assault.

Table 4 : Basis and stage of case withdrawal
Prosecutor's/ magistrate's assessmentVictim reluctance(Total)
Source: Australian Institute of Criminology, Sexual assault DPP study, 1 July 1999 to 30 June 2001 [computer file]
Total cases withdrawn 29 24 (53)
Prior to presentment of indictment 19 19 (38)
After indictment was filed 10 5 (15)
Table 5 : Statistically significant differences between cases that proceeded and cases that were withdrawn
VariablesCases proceededCases withdrawn
* p<.05 chi-square="" test="" of="" significance="">
** This finding is likely to reflect the high proportion of NT and WA cases in the sample
Source: Australian Institute of Criminology, Sexual assault DPP study, 1 July 1999 to 30 June 2001 [computer file]
Victim expressed non-consent
No 26 31 27 51
Yes* 59 69 26 49
Victim injury
No 35 40 31 56
Yes* 52 60 22 42
No 17 20 23 43
Yes* 70 81 30 57
Severity of assault
No threat, force, weapon, injury 11 13 18 34
Some threat, force, injury or weapon use* 76 87 35 66
Additional evidence
No 46 53 39 74
Yes* 41 47 14 26
Defendant race/ethnicity**
Caucasian 25 32 24 53
Non-Caucasian* 54 68 21 47
Relationship to victim
Known 61 70 45 87
Stranger* 26 30 7 14

The finding that cases involving non-Caucasian defendants were significantly more likely to proceed should be interpreted with caution; it does not necessarily indicate that prosecutors' decisions were based on discriminatory judgments. The sample may not be representative of all sexual assaults referred for prosecution over the reference period, and the results are likely to be skewed by the high proportion of NT and WA cases. Further research is required to verify the reliability of this finding.

The finding that cases involving strangers were significantly more likely to proceed is consistent with earlier research. In this sample, three-quarters of the defendants were known to victims, and almost 42 per cent of defendants were current partners, former partners or family members. Table 6 charts the outcomes for individual defendants (not cases, as in the previous analysis), using a more fine-grained categorisation of relationships. Compared to outcomes for all defendants, fewer strangers had their cases withdrawn, were acquitted at trial or negotiated charges in exchange for a guilty plea, while partners and former partners were more likely to have their cases withdrawn and less likely to be found guilty.

Almost half of the cases were withdrawn due to the victim's reluctance to proceed (n=24), and the majority of these involved current partners (n=7), former partners (n=7), and other known defendants (n=6). Two cases involved family members and two involved strangers. Assuming that some of the discontinued cases involved defendants who were guilty of the charges against them, a number of victims were likely to return to social or domestic contexts where there was a possibility of future contact with the defendant, or the potential for repeat victimisation.

Table 6 : Outcomes for individual defendants by relationship to victim
RelationshipTotal knownWithdrawnPleaded guiltyAcquitted/dismissedFound guiltyCharge negotiation
Source: Australian Institute of Criminology, Sexual assault DPP study, 1 July 1999 to 30 June 2001 [computer file]
Stranger 35 100 7 20 14 40 4 11 9 26 5 14
Current partner 23 100 11 48 8 35 4 17 0 0 5 22
Former partner 17 100 9 53 3 18 4 24 1 6 2 12
Family 22 100 5 23 8 36 5 23 3 14 3 14
Other known 52 100 21 40 14 27 11 21 4 8 10 19
All defendants 149 100 53 36 47 32 28 19 17 11 25 17

Bivariate analyses do not indicate which of the seven variables (see Table 5) are most important in predicting case withdrawal, or whether they retain their importance when other variables of interest are controlled for. To answer these questions, logistic regression analyses were conducted using six of the seven statistically significant variables. 'Severity of assault' was excluded because it is not independent from the other variables. Information on 121 cases was analysed.

These analyses showed that, on their own, none of the variables predicted the likelihood of a case proceeding. However, the decision to proceed was reliably predicted by an interaction between 'force' and 'non-consent'. That is, cases were significantly more likely to proceed if the defendants used force and the victims actively expressed non-consent. This is understandable, as the level of force tends to be commensurate with the level of resistance, and both factors help to establish the mental element of the offence; that is, that the defendant knew the victim was not consenting, but intended to have sex regardless.

In this multi-jurisdictional sample of adult sexual assault cases, the exercise of prosecutorial discretion accounted for a relatively large degree of case attrition, with 38 per cent of cases in the sample withdrawn from prosecution. The second major point of attrition is observed in conviction rates: overall, 44 per cent of cases resulted in a conviction, but this figure encompasses a sizeable number of cases finalised by way of a guilty plea. This points to prosecutors' willingness to negotiate concessions on charges and penalties in exchange for the defendant agreeing to plead guilty, rather than risk an acquittal.

In much of the literature, attrition at the prosecution stage is couched in terms of insufficient evidence. Evidentiary sufficiency is influenced by the presence of legal and extralegal factors that are seen to reduce uncertainty over the prospects of conviction, so that only the strongest and most serious cases proceed. Thus, the prosecution's case is seen to be strengthened when the victim actively showed non-consent or sustained injuries during the attack, or when the defendant used force. The current findings can therefore be seen to support the view that the likelihood of cases proceeding increases significantly where prosecutors are more certain about the prospects of success, or in more serious cases (Albonetti 1987; Spohn & Holleran 2001).

At the same time, it should be acknowledged that prosecutors' decisions are not necessarily objective, but it is difficult to measure the effect of embedded assumptions about sexual assault and gender relations on decision-making. For example, different perspectives can be taken on the finding that stranger cases were significantly more likely to proceed. On the one hand, a critical view of prosecutorial decision-making suggests that the victim-defendant relationship is extraneous to the legal elements of the case and that decisions to discontinue cases involving prior relationships may be underpinned by stereotypical assumptions about gender relations and the nature of violence in relationships (Stanko 1982). The current study cannot rule out this possibility. On the other hand, from the prosecutor's point of view the victim-defendant relationship is relevant in a non-legal sense. It is understandable that experienced prosecutors assess the prospects of conviction by considering the relationship in combination with evidentiary factors, particularly because lack of consent may be more difficult to prove in a one-on-one contest between persons in an established relationship.

It is also worth noting that victims in cases involving prior relationships often request that the case be withdrawn. From one perspective, victims should have little say concerning prosecutorial decisions, but it is understandable that, as those most directly affected by the sexual assault and by the publicity and outcomes of a trial, they are likely to have strong views on case decisions. However, decisions not to proceed can result in public criticism of the effectiveness of the criminal justice system, especially given fears that defendants who were guilty of the charges may reoffend. Prosecutors are not precluded from pursuing these cases, but the goal of avoiding uncertainty makes it unlikely that they will do so (Spohn, Beichner & Davis-Frenzel 2002).

It must be acknowledged that the victim's reluctance to proceed provides the prosecutor with a legal rationale for discontinuance and the factors that influence victims' preferences are not well understood. There is some evidence that victims show more willingness to pursue prosecution in the presence of evidentiary factors that increase the prospects of conviction (Kerstetter 1990) and that criminal justice officials sometimes manipulate victims' choices in line with their assessments of the prospects of conviction (Kerstetter & van Winkle 1990). A victim may perceive a prosecutor's advice that conviction is highly unlikely as an invitation not to proceed. This form of 'discouragement' may be intentional or unintentional and perhaps even altruistic in some instances. For example, a victim who voluntarily withdraws from prosecution is spared being told that a jury would not find her a credible witness. However, prosecutors who pre-empt the views of the jury on the basis of various sociodemographic, personal or other extralegal variables may exclude particular types of victims from justice. They may then fail to challenge and expand the bounds of convictability by dealing with actual juries to learn how to win believable but risky cases. Above all, the results of this study suggest that there is a pressing need for further research about the way that human interactions and organisational, systemic and social variables influence both prosecutors' and victims' decisions.

This paper summarises findings from a multiple methods study of prosecutorial decisions in adult sexual assault cases (Lievore 2004). Funding for this project is provided through the National Initiative to Combat Sexual Assault - an Australian Government initiative administered by the Office of the Status of Women. The author is grateful to the directors and staff of participating DPPs for their support and assistance. Appreciation is also extended to Tamara Blakemore for her perceptive comments and skills in data analysis.


  • 1 The parties to sexual assaults are referred to by different terms throughout the criminal justice process and by different sectors. For simplicity, this paper refers to victims and defendants. Feminine terms are used to refer to victims and masculine terms to defendants. While both women and men can be victims and perpetrators of sexual assault, 82 per cent of all sexual assault victims who came to the attention of Australian police in 2003 were female (ABS 2004). Responses to the Women's Safety Australia Survey indicated that 99 per cent of offenders were male (ABS 1996).


  • Albonetti CA 1987. Prosecutorial discretion: the effects of uncertainty. Law & society review 21(2): 291-313
  • Australian Bureau of Statistics (ABS) 2004. 2003 recorded crime - victims, Australia (cat no 4510.0) Canberra: Australian Bureau of Statistics
  • --- 1997. Australian standard offence classification 1997 (cat no 1234.0) Canberra: Australian Bureau of Statistics
  • --- 1996. Women's safety Australia (cat no 4128.0) Canberra: Australian Bureau of Statistics
  • Briody M 2002. The effects of DNA evidence on sexual offence cases in court. Current issues in criminal justice 14(2): 159-181
  • Bryden DB & Lengnick S 1997. Rape in the criminal justice system. The journal of criminal law & criminology 87(4): 1194-1384
  • Frohmann L 1998. Constituting power in sexual assault cases: prosecutorial strategies for victim management. Social problems 45(3): 393-407
  • Harris J & Grace S 1999. A question of evidence? Investigating and prosecuting rape in the 1990s. London: Home Office
  • Kerstetter W 1990. Gateway to justice: police and prosecutorial response to sexual assaults against women. Criminology 81(2): 267-313
  • Kerstetter W & van Winkle B 1990. Who decides? A study of the victim's decision to prosecute in rape cases. Criminal justice and behavior 17(3): 268-283
  • Kingsnorth RF, MacIntosh RC & Wentworth J 1999. Sexual assault: the role of prior relationship and victim characteristics in case processing. Justice quarterly 16(2): 275-302
  • Lievore D 2004. Prosecutorial decisions in adult sexual assault cases: an Australian study. Canberra: Australian Government Office of the Status of Women
  • Refshauge R 2002. Prosecutorial discretion - Australia. In G Moens & R Biffot (eds) The convergence of legal systems in the 21st century: 353-390. Brisbane: CopyRight Publishing
  • Spohn C, Beichner D & Davis-Frenzel E 2002. Prosecutorial justifications for sexual assault case rejection: guarding the 'gateway to justice'. Social problems 48(2): 206-235
  • Spohn C & Holleran D 2001. Prosecuting sexual assault: a comparison of charging decisions in sexual assault cases involving strangers, acquaintances and intimate partners. Justice quarterly 18(3): 651-688
  • Stanko EA 1982. Would you believe this woman? Prosecutorial screening for 'credible' witnesses and a problem of justice. In NH Rafter & EA Stanko (eds) Judge, lawyer, victim, thief: women, gender roles, and criminal justice: 63-82. Boston: Northeastern University Press
Dr Denise Lievore is a research analyst in the justice and crime analysis program at the Australian Institute of Criminology

Related links