This feature comprises an essay written by an experienced injury epidemiologist, and three commentaries representing perspectives from psychiatry and population health, sociology and forensic pathology. At issue is an apparent decline during the new millennium in the quality of suicide statistics in the USA. Perpetually stigmatised and often socially condemned, suicide is highly prone to under-reporting. The ongoing opioid and suicide epidemics and the recent COVID-19 pandemic have overwhelmed already-stressed and under-resourced medical examiners and coroners (ME/Cs), who must distinguish suicide from other external manners of death (eg, homicide, ‘accident’ and undetermined intent). The frequent absence or paucity of corroborative evidence, such as an authenticated suicide note or a well-documented psychological/psychiatric history, makes the medicolegal determination of suicide especially fraught when ME/Cs must discern intent among decedents of drug intoxication, as compared with deaths due to less ambiguous methods (eg, firearms, jumping and hanging). A common result is the default determination of ‘accident’, especially when investigative resources are limited. Accurate measurement is essential for optimising mental health treatment and suicide prevention. Bias and differential bias implicit in suicide detection would be mitigated by an expansion of the self-injury mortality domain, which emphasises antemortem behaviours of decedents rather than postmortem inference of their intentionality towards death.