Self-harm and suicide is more prevalent within the prison environment than in community samples, with those in the first weeks of imprisonment at greatest risk. Descriptions and evaluations of static risk factors (e.g. mental health diagnosis) dominate the empirical literature with few dynamic (e.g. defeat) and protective factors (e.g. resilience) evaluated. Additionally, current research is largely atheoretical and the integration of existing knowledge into a unifying model may improve the predictability of assessment. In the current research Williams and Pollock’s cry of pain model provided the template for assessing predictors of self-harm or suicide. For three months, all new arrivals at a local prison were invited to complete baseline questionnaires to assess factors derived from the cry of pain model. It was hypothesised that the factors derived from the model (perceived stress, defeat, entrapment and absence of rescue factors) would be predictive of self-harm and suicide risk and would distinguish prior self-harmers from non self-harmers. Two hundred and seventy prisoners participated in the study. Prisoners with active psychosis and non-English speakers were excluded. All participants were followed up for four months for instances of self-harm. Eighteen participants engaged in self-harm during this period. The hypotheses derived from the model were supported in the prediction of future engagement in self-harm in prison and had some support in identifying those who engaged in previous self-harm and those at risk of suicide. Additional research is needed to confirm the factor structure of defeat and entrapment and the presence of ‘scripts’ as relevant factors in the cry of pain model. The 3 implications for practice are discussed including the identification of patterns of risk linked with self-harm and suicide. The measures utilised in the study were shown to be largely valid within this population. Methodological limitations are discussed together with their implications for future research.