BACKGROUND: In randomized, controlled trials, antidepressant medications have been shown to reduce the duration of major depressive episodes and to reduce the frequency of relapse during long-term treatment. The epidemiological impact of antidepressant use on episode duration and relapse frequency, however, has not been described. METHODS: Data from two Canadian general health surveys were used in this analysis: the National Population Health Survey (NPHS) and the Canadian Community Health Survey (CCHS). The NPHS is a longitudinal study that collected data between 1994 and 2000. These longitudinal data allowed an approximation of episode incidence to be calculated. The cross-sectional CCHS allowed estimation of episode duration. The surveys used the same sampling frame and both incorporated a Short Form version of the Composite International Diagnostic Interview. RESULTS: Episodes occurring in antidepressant users lasted longer than those in non-users. The apparent incidence of major depressive episodes among those taking antidepressants was higher than that among respondents not taking antidepressants. Changes in duration and incidence over the data collection interval were not observed. CONCLUSIONS: The most probable explanation for these results is confounding by indication and/or severity: members of the general population who are taking antidepressants probably have more highly recurrent and more severe mood disorders. In part, this may have been due to the use of a brief predictive diagnostic interview, which may be prone to detection of sub-clinical cases. Whereas antidepressant use increased considerably over the data-collection period, differences in episode incidence and duration over time were not observed. This suggests that the impact of antidepressant medications on population health may have been less than expected.