The Short-Term Assessment of Risk and Treatability (START) guides assessment of potential adverse outcomes. Assessment is recommended every 3 months but there is no evidence for this interval. We aimed to inform whether earlier reassessment was warranted. We collated START assessments for N = 217 adults in a secure mental health hospital, and subsequent aggressive, self-harm, self-neglect and victimization incidents. We used receiver operating characteristic analysis to assess predictive validity; survival function analysis to examine differences between low-, medium-, and high-risk groups; and hazard function analysis to determine the optimum interval for reassessment. The START predicted aggression and self-harm at 1, 2 and 3 months. At-risk individuals engaged in adverse outcomes earlier than low-risk patients. About half warranted reassessment before 3 months due to engagement in risk behaviour before that point despite a low-risk rating, or because of non-engagement by that point despite an elevated risk rating. Risk assessment should occur at appropriate intervals so that management strategies can be individually tailored. Assessment at 3-month intervals is supported by the evidence. START assessments should be revisited earlier if risk behaviours are not prevented; teams should constantly re-evaluate the need for restrictive practices.