In this literature review, the failure of a disease model of alcoholism based upon loss of control and craving phenomena to receive support is noted, as is the robustness of a model rooted in psychological learning theory. The viability of controlled drinking interventions based upon learning theory principles is demonstrated. Several predictor variables which emerged from studies using gamma alcoholics are identified: (1) low severity of dependence symptoms; (2) age 40 or less; (3) regular employment (favors a controlled drinking outcome); and (4) Alcoholics Anonymous involvement (favors an abstinent outcome). Possible learning theory explanations of these predictor variables are presented. Physical health/safety issues concerning controlled drinking research are also discussed. An apparently safe operational definition of alcohol consumption for future controlled drinking research of no more than 12.75 oz. ethanol equivalent/week, up to 2-5 oz. ethanol equivalent/drinking day is proposed. Other conclusions and suggestions for further research are offered. Seven pages of references conclude the report. (Author/NB)