The purpose of this study was to evaluate options available for the treatment of neuroleptic malignant syndrome. We conducted a computer search of English-language literature using a Medline database and review of bibliographies of identified articles. Sixty-four case reports containing sufficient data about patients' demographics and clinical response to therapy to allow evaluation were combined with three previously unreported cases. The mean time to clinical response was 6.8 days when supportive measures alone were used in the treatment of neuroleptic malignant syndrome. The addition of dantrolene (response time, 1.15 days) or bromocriptine (response time, 1.03 days) significantly shortened the time to clinical response. We conclude that the addition of either dantrolene or bromocriptine to conservative measures decreases the time to clinical response. A randomized trial would address potential case selection biases.