BACKGROUND There has been a great deal of research on the comorbidity between alcohol abuse or dependence (AAD) and major depression (MD). However, it is unclear whether AAD increases the risk of MD or vice versa. OBJECTIVE To examine the associations between AAD and MD using fixed-effects modeling to control for confounding and using structural equation models to ascertain the direction of causality. METHODS Data were gathered during the course of the Christchurch Health and Development Study, a 25-year longitudinal study of a birth cohort of children from New Zealand (635 boys, 630 girls). METHODS General community sample. METHODS The analysis was based on a sample of 1055 participants with available data on AAD and MD at ages 17 to 18, 20 to 21, and 24 to 25 years. METHODS Symptom criteria for AAD and MD from the DSM-IV at ages 17 to 18, 20 to 21, and 24 to 25 years as well as measures of life stress, cannabis use, other illicit drug use, affiliation with deviant peers, unemployment, partner substance use, and partner criminality at ages 17 to 18, 20 to 21, and 24 to 25 years. RESULTS There were significant (P < .001) pooled associations between AAD and MD. Controlling for confounding factors using conditional fixed-effects models and time-dynamic covariate factors reduced the magnitude of these associations, but they remained statistically significant. Structural equation modeling suggested that the best-fitting causal model was one in which AAD led to increased risk of MD. CONCLUSIONS The findings suggest that the associations between AAD and MD were best explained by a causal model in which problems with alcohol led to increased risk of MD as opposed to a self-medication model in which MD led to increased risk of AAD.