Despite numerous pharmacological investigations of intracranial self-stimulation (ICSS), the substrates of this behavior have yet to be completely understood. In view of the likelihood that inadequate methodology has hindered the quest for these substrates, the present review was undertaken. Criteria for ICSS methodology should include not only the ability to discriminate reward from gross performance deficit, but also adequate capacity (ability to generate experimental data at a reasonable rate). For numerous reasons, bar-pressing on a continuous reinforcement schedule fails the first criterion despite its ease and rapidity. The use of partial reinforcement schedules may alleviate some of these shortcomings. Analysis of drug-induced response decrement patterns can discriminate gross motoric incapacity from other variables, although the question of subtle response maintenance deficits remains to be answered. Measurements of response rates using alternative operants do not differentiate reward and performance adequately. More promising, "rate-free" measures using locomotion as an operant include the two-platform method of Valenstein and the "locus of rise" method. Comparison of drug effects on ICSS with those on alternate tasks are fraught with pitfalls including the problems of assuring equivalent rates and patterns of responding. The use of differential electrode placements is ideally suited for neurochemically well-characterized drugs, particularly if "double dissociations" can be established during studies of multiple placements. Presentation of different current intensities or frequencies permits the compilation of rate-intensity functions, and drug-induced shifts in these functions have considerable analytical power. Self-regulation of current intensity constitutes a powerful tool that has yet to realize its full potential in the pharmacological study of ICSS. Extensive studies involving self-regulation of stimulation duration ("shuttlebox" studies) suggest that this method may be highly versatile despite several practical difficulties. It is concluded that at least six of these methods appear to do a reasonable job of excluding gross performance deficit. However, the possible influences of other factors, such as subtle response maintenance deficit, incentive or arousal, remain to be resolved in view of the multifactorial nature of ICSS. Multiple tests for ICSS drug or lesion studies are advocated whenever feasible, as no single test appears capable of resolving all theoretical complexities.