To compare the efficacy and safety of the intramuscular formulations of ziprasidone, olanzapine, and aripiprazole in treating agitation. The pivotal registration trials were accessed by querying on-line literature and clinical trial databases. Pharmacovigilance data and posters were requested from the manufacturers. No date or language constraints were applied. Nine double-blind, randomized, controlled clinical trials were identified. Number needed to treat (NNT) for response to treatment for agitation and number needed to harm (NNH) for extrapyramidal effects were calculated from the study reports. Additional safety outcomes subject to NNH analysis were obtained from product labeling. Using the a priori definitions of response at 2 hours after the first injection, NNT for response versus placebo (or placebo equivalent) in treating agitation for the pooled data at the recommended dose of ziprasidone 10-20 mg was 3 (95% CI = 2 to 4), for olanzapine 10 mg was 3 (95% CI = 2 to 3), and for aripiprazole 9.75 mg was 5 (95% CI = 4 to 8). Treatment-emergent adverse events occurring during the pivotal trials revealed statistically significant NNH versus placebo (or placebo equivalent) for aripiprazole for headache (NNH = 20, 95% CI = 11 to 170) and nausea (NNH = 17, 95% CI = 11 to 38), for ziprasidone in the treatment of headache (NNH = 15, 95% CI = 8 to 703), and for olanzapine in treatment-emergent hypotension (NNH = 50, 95% CI = 30 to 154). Olanzapine and aripiprazole had a more favorable extrapyramidal side effect profile compared to haloperidol. (There was no haloperidol treatment arm in the ziprasidone studies.) Although the lowest NNT, and hence strongest therapeutic effect, was seen for the studies of ziprasidone and olanzapine as opposed to aripiprazole, head-to-head controlled studies directly comparing these 3 agents are needed.