The objective of this article was to estimate the risk of discontinuation due to adverse events in trials of orlistat, sibutramine and rimonabant. Medline, EMBASE, the Cochrane controlled trials register and reference lists of identified articles were searched from 1990 to May 2008. All randomized placebo-controlled trials of 12-24 months of duration on adults using licensed doses were included. Studies/study arms were excluded if they evaluated weight maintenance after weight loss. Trials were identified, subjected to inclusion and exclusion criteria and reviewed. Data on participants, interventions and discontinuation were extracted and trials rated for quality based on established criteria. A random effects model was used to estimate pooled risk ratios, risk differences and number needed to harm (NNH). A total of 28 trials met the inclusion criteria (16 orlistat, 7 sibutramine and 5 rimonabant). The risk ratios for discontinuation due to adverse events were significantly elevated for rimonabant (2.00; 1.66-2.41) and orlistat (1.59; 1.21-2.08), but not sibutramine (0.98, 0.68-1.41). Compared with placebo, the risk difference was the largest for rimonabant (7%, 5-9%; NNH 14, 11-19), followed by orlistat (3%, 1-4%; NNH 39, 25-83), while no significant difference was seen for sibutramine (0.2%, -3 to 4%; NNH 500). The most common adverse events leading to withdrawal were gastrointestinal for orlistat (40%) and psychiatric for rimonabant (47%). Corresponding information was unavailable for sibutramine. In conclusion, available weight loss drugs differ markedly regarding risk of discontinuation due to adverse events, as well as in underlying causes of these events. Given the large number of patients eligible for treatment, the low NNH for rimonabant is a concern.