Data from a variety of sources suggest the possibility of significant errors in lifetime recall of psychiatric illness. Remote episodes of illness may often be forgotten, especially among patients with milder or less recurrent illness or those who received no treatment. These errors in recall may lead to significant underestimation of true lifetime prevalence and may produce the appearance of large temporal increases in risk in the absence of any true change. When assessing lifetime history, cross-sectional surveys may differentially underestimate prevalence in the elderly and in those with less severe or recurrent illness. While lifetime risk of depression and other psychiatric disorders may not be increasing as rapidly as suggested by cross-sectional surveys, the overall impact of such disorders over the life span may be significantly greater than previous community surveys have indicated.