A meta-analysis of 67 published randomized controlled clinical trials comparing selective serotonin reuptake inhibitors (SSRIs) and tricyclic antidepressants (TCAs), which measured discontinuation rates for side effects and lack of efficacy, was performed. All multiple publications and trials using non-TCA comparators were excluded. Ten studies were placebo controlled; these were analysed separately. Overall, the difference in withdrawals due to side effects of SSRIs and TCAs was -4.5% (p = 0.0004) and that due to lack of efficacy was 0.1% (p = 0.86). In the placebo-controlled trials the differences between the two groups were -7.9% and -0.1% (p = 0.06 and 0.96), respectively. These results demonstrate that SSRIs have a significant and clinically important advantage over TCAs with respect to tolerability, whereas efficacy is similar. Treatment failure due to poor compliance can increase health-care costs: therefore, in selecting an antidepressant for the first-line treatment of major depressive disorders, the risks, benefits and costs of each type of treatment need to be critically evaluated.