A meta-analysis of the efficacy and tolerability of selective serotonin reuptake inhibitors (SSRIs) against nonselective and noradrenergic reuptake inhibitors (mainly tricyclic antidepressants, TCAs) in depressed inpatients was carried out. Twenty-five double-blind studies provided data on relative efficacy which was determined by a summary variance-weighted mean effect size calculated from the difference in the reduction in mean Hamilton Depression Rating Scale (HDRS) scores for the two antidepressants in each study. Twenty-three studies provided data on dropouts and relative tolerability which was determined as the variance-weighted pooling of the relative risk (RR) of dropout for all reasons and for adverse effects from each study. TCAs were significantly more effective than SSRIs (effect size = -0.23, 95% CI -0.40 to -0.05, P = 0.011), although sensitivity analyses by analysing larger studies (> 100 patients) and those providing complete data reduced the advantage to TCAs to a trend (P < 0.10). When the TCAs were grouped into those with dual action on 5-hydroxytryptamine (HT) and noradrenaline reuptake (clomipramine and amitriptyline) and those with predominantly noradrenaline reuptake (imipramine, desipramine and maprotiline), only the dual action TCAs had greater efficacy than SSRIs (effect size = -0.30, 95% CI -0.54 to -0.05, P = 0.017). When TCAs were considered individually, only amitriptyline was significantly more effective than comparator SSRIs (effect size = -0.37, 95% CI -0.67 to -0.07, P = 0.015). More patients overall discontinued treatment on TCAs than on SSRIs (29.0% vs. 25.5%), although this did not reach statistical significance (RR = 0.88, 95% CI 0.75 to 1.03, P = 0.121). However, significantly more patients stopped treatment due to adverse effects on TCAs compared to SSRIs (14.2% vs. 9.1%, RR = 0.66, 95% CI 0.50 to 0.87, P = 0.003) with no difference in discontinuations due to treatment failure (10% vs. 11.6%, RR = 1.13, 95% CI 0.84 to 1.51, P = 0.42). This meta-analysis suggests that at least some TCAs may be more effective than SSRIs in depressed inpatients, with there being the strongest evidence for amitriptyline. A possible explanation is that this is related to a dual action in inhibiting both 5-HT and noradrenaline reuptake. In agreement with previous meta-analyses, TCAs appear less well tolerated than SSRIs, although the absolute risk difference for discontinuation due to adverse effects (4.9%, 95% CI -8.1 to -1.7%) is of uncertain clinical significance.