Fifty-six prevalence surveys of tardive dyskinesia (TD) in neuroleptic-treated patients are reviewed, yielding an average prevalence of 20% as compared with 5% prevalence of "spontaneous" dyskinesia in 19 samples of untreated individuals. Reported prevalence has increased during the past two decades and is not attributable simply to the more widespread use of rating scales. Controlled studies of a variety of risk factors are reviewed and suggest that advancing age and, to a lesser extent, female sex are the two variables most consistently associated with increased prevalence. There are relatively few data supporting the assumption that the risk of TD development continues to increase with increasing neuroleptic exposure, CNS dysfunction, or exposure to antiparkinsonism medication.