The prevalence of minor psychiatric morbidity (MPM) was investigated in 523 adults presenting to the casualty section of an Australian public hospital. Relationships between subject characteristics and the level of MPM were also examined, together with the acceptability of an offer of short-term intervention and the effectiveness of this intervention in reducing distress. Threshold morbidity on initial presentation to casualty was 41% (using the GHQ-12 with the customary 1/2 cut-off point) and the probable prevalence was estimated to be about 27%. The intervention was initially offered to a random sample of those identified as being 'under stress' (GHQ of three or above), but as there was a 90% rejection rate, it was offered to all symptomatic subjects. Overall, 26% accepted our offer of intervention. Those who accepted intervention and those who rejected it did not differ in their improvement in MPM during the subsequent six months. However, these groups improved slightly more than the group that was eligible for intervention but to whom it was not offered. Forty-seven per cent of subjects who were 'under stress' on initial presentation to casualty fell below the GHQ-12 threshold at the six-month follow-up; this compared with a shift of only 26% in the opposite direction (i.e., non-stressed to stressed). Regression to the mean offered the most parsimonious explanation for these GHQ changes.