Optimal cut-off points on the Severity of Dependence Scale (SDS) indicative of clinically significant dependence have been determined for a range of substance types. This study aims to determine a cut-off point on SDS that discriminates between the presence and absence of a DSM-IV diagnosis of alcohol dependence. A structured interview was administered to 90 alcohol users in Sydney, Australia. Receiver Operating Characteristic curve analysis confirmed the utility of the SDS-alcohol for characterising and diagnosing persons with respect to their alcohol-dependent status to an accuracy of 85%. A SDS score of 3 or above was determined as optimal for characterising alcohol dependence. Evidence is also provided confirming that the SDS-alcohol is a valid, reliable uni-dimensional scale for measuring alcohol dependence. It has been demonstrated that the SDS-alcohol can be used to characterise an individual's alcohol-dependent status. A cut-off value for SDS-alcohol provides additional meaning and value to the scale for clients and clinicians and will enable researchers to characterise the prevalence of alcohol dependence in their target populations.