Second order schedules of IV cocaine reinforcement in rats provide a reliable method for evaluating the effects of conditioned stimuli on cocaine-seeking behaviour, and for measuring the motivational aspects of cocaine reinforcement. In the procedure established here, each infusion of cocaine (0.25 mg/infusion) was initially made contingent on a lever press and was paired with a 20-s light conditioned stimulus (CS). When rats acquired stable rates of cocaine self-administration, the response requirement for cocaine was increased progressively to a second-order schedule of the type FI15 min(FR10:S), whereby the IV cocaine infusion was self-administered following the completion of the first FR10 responses (and CS presentation) after a 15-min fixed interval (FI) had elapsed. Evaluation of the animals' responding during the first, drug-free interval of each daily session provided a measure of cocaine-seeking behaviour, independent of other pharmacological effects of the self-administered drug. Thus, a dose-response study (dose range: 0.083, 0.25 and 0.50 mg/infusion) revealed that responding under this schedule during the initial, drug-free interval changed monotonically with dose, whereas an inverse relationship between cocaine dose and response level tended to appear during the rest of the session, after rats had self-administered the drug. Responding under this schedule was also shown to occur under the control of the CS, which had acquired conditioned reinforcing properties. Thus, a decrease in responding and an increase in the latency to initiate responding followed the omission of the CS for 3 consecutive days. In addition, extinction of cocaine-seeking behaviour was slower when contingent CS presentations occurred compared to extinction when the CS was not present. Furthermore, the reinstatement of responding for cocaine, which followed a brief period of non-contingent CS presentations, was retarded when this conditioned reinforcer had been extinguished together with cocaine. Finally, cocaine-seeking behaviour decreased markedly for the first 6 h that followed a 12-h period of continuous access to cocaine, when compared to responding 6 h after a 90-min session of limited access to the drug. Responding subsequently increased to baseline levels within 72 h. These results emphasise the utility of second-order schedules for studying drug-seeking behaviour and the importance of drug-associated cues in maintaining such responding for cocaine.