Regulatory control of drugs with abuse liability is an important component of drug control policy and is believed to help prevent nonmedical use. To be maximally effective, this requires a scientific assessment of abuse liability of drugs considered for regulatory control. These assessments have relied extensively on laboratory-based animal and human testing, but also utilize information from clinical trials, actual abuse and other sources. Here, we discuss recommendations and guidelines that have been proposed for abuse liability assessment and describe important review papers and conference proceedings that have addressed this matter, focusing primarily on drugs with medical usefulness. Historically, there is substantial consensus about how to approach abuse liability evaluation of drugs with actions similar to those of abused opiates, stimulants, depressants, and to a somewhat lesser extent, cannabinoids and hallucinogens, and much of what has been recommended for abuse potential assessment in the past remains valid and useful. On the other hand, novel CNS-active medications which cannot be readily classified with these traditional drugs of abuse are increasingly under development. In addition, advances in the science of abuse liability assessment need to be incorporated into future guidelines and recommendations on this subject. Developers of new medications need guidance on how to utilize scientific research to maximize therapeutic benefit while minimizing risk for abuse. Thus, another goal of this review has been to identify areas where critical thinking and new guideline development are needed.