Since the application of the Obsessive Compulsive Drinking Scale (OCDS) has been reported to be problematic when used to measure alcohol craving in longitudinal studies, we examined the following questions: (1) Is it possible to skip problematic quantity items? (2) Is the score calculation rule using the higher value of item pairs necessary? (3) Can the shortened version of the OCDS be applied alternatively? We examined two samples including a total of 355 alcohol-dependent patients: a multi center study sample (n=149) and a validation control sample (n=206). Neither an advantage of the score calculation rule nor the necessity of including items regarding alcohol consumption could be demonstrated. The exclusion of consumption items lead to a clear, stable 2-factor structure with a maximum stability (.81-.91). Retest-reliability ranged from r(tt)=.73 to r(tt)=.76 at an average time interval of 5 weeks. Concerning stability (.68-.81) and reliability (r(tt)=.76), the short version turned out to be equivalent. The short version of the OCDS seems to be sufficient. If different effects on cognitive and behavioral levels are expected, the 12-item version without the quantity items should be applied.