We reviewed 285 treatment studies involving more than 3,000 patients with neuroleptic-induced tardive dyskinesia (TD). Neuroleptic withdrawal is found to reverse dyskinesia in about 37% of patients. There is no satisfactory treatment for persistent TD. Although neuroleptics are significantly superior to most other methods of treatment in suppressing signs of dyskinesia, the safety of their long-term use in dyskinetic patients remains to be demonstrated. Putative gamma-aminobutyric acid (GABA)-ergic drugs and noradrenergic blockers deserve careful study. A strategy for determining biochemical and pharmacologic subtypes of TD appears promising. The value of the available cholinergic agents in the treatment of TD is uncertain. Caution is warranted in interpreting "positive" results with a number of other drugs, which might act as placebos or as nonspecific sedatives. Anticholinergic drugs are generally not recommended for treating dyskinetic patients. Current theories of the pathophysiology of TD may need revision. Drug-free periods do not seem to prevent TD. Antipsychotic drugs without neuroleptic side effects should be developed.