This article presents an overview of the definition and measurement of "craving" as it is applied to drug and alcohol abuse research. Examples of craving measures are described and organized in terms of whether they structure craving as a single factor or as a multifactorial construct. The application of craving scales in cue-conditioning studies, in medication treatment trials, and as outcome measures in other treatment studies is considered. It is concluded that no single craving scale, or type of scale, has unequivocal support, because few studies have administered more than one scale to the same population. It is suggested that our understanding of craving will be advanced by designing studies that compare and contrast several craving scales within the same study. Likewise, craving is only one possible cause of relapse. Future studies should compare several alternate relapse mechanisms within the same study.