The abuse liability of a drug is closely related to its ability to maintain self-administration behavior in laboratory subjects. But how do researchers gauge the reinforcing value of a self-administered drug in the preclinical laboratory? One approach is to determine the "preference" for that drug, that is, the allocation of behavior to drug taking, when alternative reinforcers are concurrently available. Careful analyses of such "choice" behavior in laboratory subjects can lead to a scientific understanding of the pharmacological and behavioral determinants of the reinforcing strength of a drug and, ultimately, to a more useful preclinical evaluation of abuse liability.