There is considerable evidence from inter-respondent and test–retest studies that the data obtained in structured life event inventories are not very reliable. Inter-respondent reliability estimates are presented in this paper for the life event reports obtained in a community survey that was designed to facilitate recall of life events. It is documented that respondents were able to recall many sorts of severe negative events with adequate reliability over a twelve-month recall period. Data are also presented to show that respondents could date event occurrence with good consistency. Sensitive events, such as involvement in criminal activities and difficulties in marriage, were not reliably reported. These results suggest that the problems of faulty memory and conceptual confusion about serious events can be overcome with careful question wording and administration procedures, but that the problem of respondent reticence about reporting sensitive events remains unresolved.