Acetazolamide is used to prevent acute mountain sickness (AMS). We assessed efficacy and harm of acetazolamide for the prevention of AMS, and tested for dose-responsiveness. We systematically searched electronic databases (until April 2011) for randomized trials comparing acetazolamide with placebo for the prevention of AMS. For each dose, risk ratios were aggregated using a Mantel-Haenszel fixed effect model. Numbers needed to treat (NNT) to benefit one subject with each dose were calculated for different baseline risks. Modes of ascent were taken as proxies of baseline risks. Twenty-four trials were included; 1011 subjects received acetazolamide 250, 500, or 750 mg day⁻¹; 854 received placebo. When climbing, median speed of ascent was 14 m/h, average AMS rate in controls was 34%, and NNT to prevent AMS with acetazolamide 250, 500, and 750 mg/day compared with placebo was 6.5, 5.9, and 5.3. When ascending by transport and subsequent climbing (speed of ascent 133 m/h) or by transport alone (491 m/h), average AMS rate in controls was 60%, and NNT with acetazolamide 250, 500, and 750 mg/day was 3.7, 3.3, and 3.0. In hypobaric chambers, median speed of ascent was 4438 m/h, average AMS rate in controls was 86%, and NNT with acetazolamide 250, 500, and 750 mg/day was 2.6, 2.3, and 2.1. The risk of paresthesia was increased with all doses. The risk of polyuria and taste disturbance was increased with 500 and 750 mg/day. The degree of efficacy of acetazolamide for the prevention of AMS is limited when the baseline risk is low, and there is some evidence of dose-responsiveness.