Operant-conditioning methods were used to train rats to reach through an opening in an operant chamber to exert forces on a silent, nearly isometric force-sensing manipulandum. The reinforcement contingency required the rat to hold forelimb force at 20-50 g for a minimum of 2 s. Once established, this 'hold-in-the-band' behavior yielded three measures of performance (time on task, number of reinforcers, variance of in-band force). Variance of in-band force was presumed to reflect steadiness of forelimb control. Acute drug effects on the three dependent variables were assessed for dose ranges of haloperidol (HAL), chlorpromazine (CPZ), clozapine (CLZ), and chlordiazepoxide (CDP). Moreover, the effects of HAL (0.5 mg/kg) and CLZ (5.0 mg/kg) were examined in a subchronic (28 day) dosing regimen. The acutely administered neuroleptics (HAL, CPZ, CLZ) produced dose-related decreases in time on task and number of reinforcers, but did not significantly affect variance of in-band force. The subchronic paradigm produced similar results. CDP did not significantly affect variance of in-band force and the 5.0 mg/kg dose produced a slight, but non-significant increase in time on task while significantly decreasing number of reinforcers; a trend opposite to that seen for the neuroleptics, which produced parallel effects on these two measures. The results suggest that the neuroleptics impaired performance by affecting the tendency to initiate responding instead of affecting the capacity to maintain steady forelimb force once a response was started.