An examination of the Brown-Harris etiological model of depressive disorder was accomplished by analyzing data from a community study of 800 middle-aged urban Swedish women. The sample included 53 onset cases of major depressive episode (DMS-III) in a 1-year period. In the analysis, interaction was defined according to the additive model employed by Brown & Harris. Occurrence of provoking agents was not significantly related to onset of depressive episode. None of the four vulnerability factors of the model showed significant interaction with provoking agents as expected from the model. One of them, lack of intimacy, increased significantly the risk of major depression in its own right, i.e. in the absence of provoking agent. Possible explanations for the major discrepancies of results between the original Camberwell study and the present one are briefly discussed.