We report a consensus conference on psychotic depression that addressed historic, classification, phenomenologic, epidemiologic, aetiologic, management and outcome issues. We were able to detail the impact of the information offered by having audience members complete questionnaires before and after the conference. The respondents indicated that the status of psychotic depression (as a separate type or as a more severe expression of depression) remains unclear; that delusions, hallucinations and severe psychomotor disturbance have high cross-sectional diagnostic weighting (while longitudinal information is of importance); that determinants include both genetic and organic factors; that most patients with this condition require admission to hospital; and that bilateral ECT is the most effective treatment. The answers to the questionnaires established areas where the audience did not modify their responses because they were already well-informed, others where their views were considerably changed (e.g. "psychotic" episodes in those with a borderline personality disorder) and others where they modified their clinical reasoning (e.g. "if a "psychotic" feature is mood-congruent then the condition must be an affective disorder" became "if the patient has an affective disorder, the psychotic feature must be mood-congruent"). Finally, we make some recommendations for future consensus conferences.