Rape offences in England and Wales garner incredibly low levels of convictions. The policing stage of a complaint experiences high levels of case discontinuances. The aim of this research is to explore police officer's decision making in rape cases and how that shapes the attrition of cases.The method employed in this study is qualitative interviews, with retired or serving police officers, with the use of vignettes. This study found that police officers prioritise and continue with cases that they predict to be most likely to reach a conviction. Such predictions are guided by assessments over the strength of the evidence in cases, informed by a range of legal and extra-legal factors. An overarching prediction employed is how likely a jury would be to convict a case. However, the participants reported that they find it incredibly difficult to accurately predict jury verdicts, making such decisions premised on a flawed logic.