We are grateful to Large & Nielssen ^{Reference Large and Nielssen1} for their interest in the report from our pilot study, ^{Reference Rahman, Gupta, While, Rodway, Ibrahim and Bickley2} but think they may have misunderstood its main purpose. We wished to investigate the ‘low risk paradox’ – the fact that in the National Confidential Inquiry data, risk is nearly always reported as low prior to suicide and homicide. ³

We agree that risk assessment in people who do not die is of interest but our study was not set up to investigate this. Equally, contradictory risk factors are of interest but our focus was on the risk assessment process itself, not on a tally of risk factors or whether they were the ‘right’ ones. Last, our study was not an investigation of the predictive utility of risk assessment. We are familiar with the low base rate problem and have written about this elsewhere. ^{Reference Kapur4}

We would wholeheartedly agree with Large & Nielssen's suggestion that assessments should be compassionate, ethical and needs-focused. However, we would take issue with their objection to risk assessment as a whole – we think it is never futile and can sometimes be life-saving.