The allied psychological drinking constructs of alcohol expectancy and drinking restraint have shown considerable promise, both independently and in combination, in understanding problematic drinking. However, previous research examining the relationship between these two constructs has only occurred in pre-dependent, convenience samples. This study examined the role of both alcohol expectancies and drinking restraint in an alcohol dependent sample. 143 DSM-IV alcohol dependent participants (93 males, 50 females) completed measures of drinking restraint (Temptation and Restraint Inventory, TRI) and alcohol expectancy (Drinking Expectancy Profile, DEP), along with quantity and frequency of consumption and dependence severity (Alcohol Dependence Scale, ADS). The results showed that although alcohol expectancy and drinking restraint do share common underlying properties, there was unique variance attributed to the prediction of dependence severity and consumption. The results also failed to replicate the drinking restraint model observed in non-clinical samples. It was concluded that alcohol expectancy and drinking restraint models appear to show superior utility in pre-dependent populations. The implications of these findings in refining social cognitive models of alcohol misuse are discussed, with a particular focus on prevention.