BACKGROUND Second-order schedules of heroin self-administration provide a method of measuring heroin-seeking behaviour independently of the effects of the drug on motor behaviour and of investigating the role of heroin-associated stimuli in such heroin-seeking behaviour. OBJECTIVE These experiments aimed to establish a second-order schedule of heroin self-administration in rats, similar to that already established in this laboratory for cocaine self-administration and to investigate the role of discrete heroin-associated stimuli in the maintenance of heroin-seeking behaviour under a second-order schedule of reinforcement. METHODS Heroin i.v. self-administration (0.04 mg/infusion) was initially contingent upon a lever press, and each infusion was paired with presentation of a 20-s light-conditioned stimulus (CS). Following acquisition of heroin self-administration, the response requirement was progressively increased so that, ultimately, responding was maintained under a fixed interval (FI) 15 min [fixed ratio (FR)5:S] second-order schedule. The effects of varying the dose of heroin (0.01 mg and 0.08 mg/infusion) and pre-treatment with the mu-opiate receptor antagonist, naloxone, on responding under a FI15(FR5:S) schedule were investigated. In addition, the role of the heroin-associated CS on responding was assessed by measuring the effects of omitting the CS during heroin-seeking behaviour and during extinction of responding, as well as the effect of CS presentation on the reinstatement of heroin-seeking behaviour following extinction. RESULTS A second-order schedule of heroin self-administration was established. There were no clear effects on heroin-seeking behaviour of increasing or decreasing the dose of heroin. Although no effect of naloxone pre-treatment was seen on heroin-seeking behaviour during the first, drug-free interval of responding, an extinction-like pattern of responding was seen in that interval during subsequent sessions. Omission of the light CS resulted in a reduction in levels of responding for i.v. heroin, indicating its role in maintaining heroin-seeking behaviour. However, under extinction conditions, response-contingent CS presentations did not affect the rate of extinction, nor did non-contingent presentations of the CS following extinction reinstate heroin-seeking behaviour. CONCLUSIONS These experiments have established a method of measuring heroin-seeking behaviour in rats by adopting a second-order schedule of i.v. heroin self-administration. The results indicate a relatively weak impact of discrete, heroin-associated cues on heroin-seeking behaviour relative to cocaine-seeking behaviour studied under similar conditions.