This study involved an attempt to develop a screening test to select animals that would learn a cocaine drug discrimination (DD) rapidly and an analysis of factors affecting rates of acquisition of cocaine DD. Rats (n = 18) were trained initially in a conditioned taste aversion (C.T.A.) procedure with cocaine (36 mgikg). Subjects were then trained to discriminate cocaine (5 mgikg) in an operant procedure, and the cocaine generalization curve was determined. Subjects were then tested for cocaine-induced hyperthermia. Animals that showed weak C.T.A. showed relatively rapid acquisition of DD (r, = -0.56, P < 0.05, two tailed). Cocaine-induced hyperthermia did not predict speed of acquisition reliably, nor did cocaine's effects on operant responding. Animals were allocated retrospectively to groups of "rapid" and "slow" discriminators. These groups did not differ in generalization EDSOs, in sensitivity to cocaine's effects on operant responding, or in hyperthermic responses to cocaine. Thus speed of DD acquisition was not dependent on individual sensitivity to cocaine. Furthermore, "rapid" and "slow" discriminators probably cannot be dissociated in terms of learning ability, since animals that acquired marked C.T.A. showed slow learning of DD. The C.T.A. data suggest that animals sensitive to dysphoric effects of cocaine are less sensitive to the drug's cueing properties and thus that the cocaine cue at 5 mgikg may be related to its euphoriant actions. It is possible that C.T.A. procedures can be used to select animals that will learn cocaine DD rapidly.