A standardized self-administration procedure in rats was used to determine the intravenous self-administration liability of graded doses of various drugs. Self-administration was reliably established with the tested addictive drugs (morphine, heroin, fentanyl and d-amphetamine), but not with the nonaddictive drugs (chlorpromazine and nalorphine). However, 1 out of 14 animals on nalorphine clearly demonstrated self-administering behavior. Self-administration was observed with delta1-tetrahydrocannabinol, but the percentage of animals (40% on the highest dose) that initiated this behavior and the amount of drug intake were low in comparison with amphetamine and narcotics. Concerning the narcotic drugs, approximate ED50 values for self-administration under the described conditions were calculated (morphine: 0.65; heroin: 0.05; fentanyl: 0.0025 mg/kg/injection). Total daily drug intake was related to the unit dose delivered per injection in that a higher drug dosage led to more drug intake. In experiments with heroin, this relationship was not caused by prior forced injections. The approximate ED50 value for amphetamine appeared to be 0.145 mg/kg/injection. Narcotic drug administration resulted in a disturbance of the patterns of food and water intake. Shortly after drug administration food intake was stimulated, followed by an increased consumption of water. The patterns of food and water intake remained disturbed in animals showing self-injecting behavior. With amphetamine both the quantity of food and the frequency of eating were reduced. These effects were observed only temporarily in animals tested without prior forced injections. The present results indicate that measuring the reinforcing efficacy of drugs under strictly defined experimental conditions provides quantitative criteria for intravenous self-administration of drugs in rats.