Withdrawal symptoms are feared by many addicts and, according to behavioural models, provide negative reinforcement for continued drug taking. Furthermore, conditioning models emphasize the role of conditioned withdrawal in precipitating relapse. A satisfactory scale for measuring withdrawal symptoms would aid the evaluation of these models and it would provide practical benefits in allowing more precise and economical use of drugs during drug withdrawal procedures. The development of such a scale, the OWS, is described. A principal components analysis showed that a single factor (severity) adequately accounted for the results. The scale has good discriminative efficiency in showing clear differences between withdrawal and post-withdrawal phases of treatment. It also discriminates clearly between those undergoing withdrawal and controls. Concurrent validity is shown by agreement with observers' ratings of withdrawal severity at high levels of severity. At lower levels, the OWS appears to be more sensitive than observers' ratings.