After "chronically equivalent" barbital and pentobarbital dosing for 5 weeks, treatments were abruptly stopped and the animals were carefully observed for signs of barbiturate withdrawal. The severity of withdrawal was assessed at preset times by counting the number of grand mal type convulsions and subjectively rating 20 additional motor, autonomic and behavioral signs including tremors, twitches, myoclonic jerks, postural disturbances and motor incoordination. Ratings achieved at peak intensity (raw scores) and their incidences were used to compute "total intensity scores" for each graded sign. For all quantitative measures, withdrawal signs were less severe for barbital than for pentobarbital, with strikingly lower (P less than .05) incidences of convulsions (6.3% vs. 100%), bizarre (hallucinatory) behavior (6.3% vs. 41.3), and death (0% vs. 100%). The withdrawal signs for barbital appeared later, developed more slowly and persisted longer than those for pentobarbital. That the onset and then peak of withdrawal signs occurred when the extents of decline from peak blood concentration of barbital and pentobarbital were similiar suggests that the time course of withdrawal might be inversely related to residual concentrations of drug, i.e., negative dose-response.