In a conti'olled stucly, inpatient violence was measured cluring placebo, high-potency (halopericlol) and lorv-potency (chlorpromazine or clozapine) neuroleptics. Some patients had a markecl increase in violent behavior rvith the mocleratelv high-dose haloperidol, but not u,ith lou,-potency neuroleptics. The authors cliscuss reasons for the inct'easecl vioience rvith halopericlol, inclucling al<athisia ancl clmg-induced behavioral toxicit.y.