OBJECTIVE: Early clinical response to antidepressant treatment is an important therapeutic goal, considering the psychological, social and economic consequences of depression. The aim of the present study was to investigate the relationship between the time course of response and the concentration of venlafaxine (V), its active metabolite O-desmethylvenlafaxine (ODV) and enantiomeric ratios V(+)/V(-) and ODV(+)/ODV(-). METHODS: Depressed inpatients ( n=35) received V orally at a fixed 300 mg daily dose. Accepted comedication included clorazepate (maximum 60 mg/day), zopiclone (maximum 15 mg/day) and low-dose trazodone (maximum 200 mg/day). Severity of depression was assessed on days 0, 4, 7, 11, 14, 21 and 28 (Montgomery and Asberg Depression Rating Scale). Blood samples were taken on day 14 and day 28 and submitted to stereoselective determination. All measurements reflected trough steady-state values. First, pattern analysis was used to provide a categorical perspective of clinical response (50% improvement from baseline depression score). Patients displaying non-response, transient response, early persistent response and delayed persistent response were compared with respect to racemic concentrations and enantiomeric ratios. Second, in a dimensional perspective, mixed-effects modelling was used to analyse severity of depression versus time curves with respect to the possible influence of concentrations and enantiomeric ratios. RESULTS: Comparison of patients with and without persistent response did not reveal any significant difference for V, ODV, V+ODV plasma levels or enantiomeric ratios. Persistent response was significantly associated with less frequent pre-study antidepressant medication and less frequent comedication with zopiclone (day 14) and clorazepate (day 28) during the study. Focus on patients with persistent response ( n=19, 54.3%) indicated that early response, first observed before day 14, was associated with significantly higher V+ODV concentration than delayed response (median 725 ng/ml versus 554 ng/ml, P=0.023). No difference was found for pre-study medication or comedication during the study. Shorter time to onset of response was significantly associated with lower V(+)/V(-) enantiomeric ratio (r(s)=0.48, P<0.05). Mixed-effects modelling of depression severity versus time curves in patients with persistent response confirmed that either higher V+ODV plasma level or lower V(+)/V(-) ratio were significantly associated with more rapid decrease of depression score (likelihood ratio tests, P=0.012 and P=0.046, respectively). CONCLUSION: Considering its modest sample size, naturalistic design and limited observation period, the present study provided preliminary indication that earlier clinical response may occur with higher V+ODV plasma level, extending previous dose-response studies. The hypothesis was also raised that exposure to a more potent noradrenergic therapeutic moiety, as reflected by a lower V(+)/V(-) ratio, may be relevant to early improvement of depression.