Measures of subjective drug craving - often defined as the experience of an intense or compelling urge or desire - may be used to predict relapse, evaluate psychological and pharmacological treatments, and test theories of addiction and craving. This review summarizes both direct self-report questionnaires and indirect behavioral, physiological and reaction time measures designed to assess craving for alcohol, amphetamines, cocaine, heroin, marijuana, and tobacco. Multi-item questionnaires have typically been based on one of four underlying conceptualizations of addiction or craving (obsessive-compulsive, approach-avoidance, multi-dimensional, intensity-frequency-duration). Most multi-item self-report questionnaires have high internal consistency, correlate significantly with single-item craving ratings, and demonstrate several aspects of construct validity. Proposed indirect or proxy measures of craving include drug dreams, speed of drug consumption, willingness to work for drug access, selection of monetary rewards over drug access, psychophysiological reactivity, and attentional bias to drug cues. These proxy measures of craving are presumed to obviate self-report biases, to be less subject to conscious self-control, and to reflect craving which the person may not be able to articulate; however, there have been too few demonstrations of their validity and they have too many practical limitations to supplant self-report measures of craving at this time.