The use of the 1980 Diagnostic Statistical Manual of Mental Disorders, Third Edition (DSM-III) alcohol abuse and depedence diagnostic system rests on the assumption that dependence is physiological and more severe than abuse. To examine the relationship of DSM-III alcohol abuse/dependence diagnoses to patients' patterns of alcohol use and the severity of their alcohol-related impairments, 920 adults (644 males, 276 females), hospitalized for alcoholism, completed the Minnesota Multiphasic Personality Inventory (MMPI) and the Alcohol and Substance Abuse Questionnaire (ASAQ). Responses to the ASAQ were classified according to Total Symptoms and were assigned a diagnosis (DSMDX) of "alcohol abuse,""alchol dependence," or "neither abuse nor dependence," based on DSM-III criteria. Five criterion measures were derived from the tests: days drunk, daily absolute alcohol consumption, problem index, general psychological impairment, and elevated MMPI scales. An analysis of the results showed significant relationships between all five of the criterion measures and both DSMDX and Total Symptoms. However, the strength of the relationship was consistently greater for the Total Symptoms than for DSMDX. Across the criterion measures, the mean proportion of variance explained by Total Symptoms was twice that explained by DSMDX. For only one criterion, general psychological impairment, did DSMDX show a significant improvement in predictability over Total Symptoms; and in this case the increment in predictability was quite small. Total Symptoms, on the other hand, showed highly significant and meaningful improvement over DSMDX for all criterion measures. Tolerance and withdrawal were only mediocre correlates of the criterion variables. These findings suggest that DSM-III diagnosis is less informative than a simple tally of symptoms. (BL)