The primary objective of this multicentre, randomized, double-blind study carried out in France was to compare the efficacy and safety of a 6-week treatment with paroxetine (20 mg/day) or mianserine (30 mg/day) in geriatric hospitalized or ambulatory patients (> or = 60 years) treated for a major depressive disorder (according to DSM III-R). A secondary objective was to discriminate those items predicting the response to an agent according to its serotoninergic or noradrenergic pharmacologic profile. The tool used for this latter purpose was the Aubin-Jouvent-Rating-Scale (AJRS) which was designed to assess the deficit of serotonin: this is a scale with 10 items, some of them regrouped into a "general" factor (irritability, sudden mood change, impatience, aggressivity) or a "depression" factor (pain, anxiety, suicidal ideas) with additional items related to sleep disorders, abnormalities in eating behavior and inability to tolerate isolation. In the perspective of this assessment, paroxetine was chosen due to its potential to inhibit serotonin re-uptake, as compared to mianserin which blocks presynaptic alpha-adrenergic receptors with negligible action on serotonin. This was a multicenter study carried out in France in 50 hospital or private practice psychiatrists. The assessment criteria included the MADRS, the AJRS, the COVI's anxiety scale, the Folstein's Mini-mental state (MMS) as well as a global assessment by the investigator at the end of the study. Safety was measured with a nondirective questionnaire, routine laboratory tests as well as a global assessment by the investigator. The primary efficacy criteria was the change in the MADRS global score. Statistical analysis included chi-square or Fisher's test as well as Student's and Wilcoxon tests for comparability at baseline, and analysis of variance for the changes in scores as during the study. A total of 116 patients was randomised (paroxetine: 54; mianserine: 62), of whom 96 completed the study (paroxetine: 43; mianserine: 53). With the exception of MADRS moderately higher in the paroxetine group, both groups were comparable at baseline. After 6 weeks of treatment, a marked improvement was recorded in both groups for all criteria except MMS; there was a consistent tendency favouring paroxetine which reached statistical significance for the COVI' scale (p = 0.001). For a given criterion, the difference paroxetine versus mianserine appeared related to the score at baseline; it was also more marked in those patients with a AJRS baseline score > or = 20 with a difference for MADRS reduction of marginal significance in favor of paroxetine (p = 0.061). As regards safety, at least one adverse event was reported in 31.5% of the patients receiving paroxetine versus 41.9% in those receiving mianserin (NS); premature withdrawal related to an adverse event was reported in 11.1% of the patients in the paroxetine group versus 12.9% in the mianserin group. No abnormality of clinical significance was reported in either group concerning laboratory tests. In conclusion, this study confirmed the therapeutic value and good safety of paroxetine as an antidepressant in geriatric populations, especially when exist a concomitant anxiety or symptoms likely to reflect a deficit of serotonine (irritability, emotional lability, restlessness, aggressivity) and to predict a good response to an agent such as this one.