OBJECTIVE To determine the predictive ability of self-report questions, physical measures, and biomarkers to detect alcohol misuse and abuse among older women. METHODS Healthy women volunteers age 60 and older who fit selection criteria were enrolled. The 135 participants were divided into nondrinkers (ND; n = 63) and drinkers (D; n = 72) based on self-reports of quantity and frequency of standard drinks consumed per month. The mean ages for the groups were 69.2 (ND) and 69.6 (D). RESULTS The best predictor was a score >0 on the T-ACE, a four-item instrument to detect alcohol abuse. Other significant predictors were: (a) behaviors: smoking, mixing over-the-counter (OTC) drugs with alcohol, heavy coffee drinking, using alcohol to sleep, and less sleep latency; and (b) biomarkers: higher mean corpuscular volume (MCV), hemoglobin (Hgb), hematocrit (Hct), and high-density lipoprotein cholesterol (HDL). The heaviest drinker subgroup had more physical stigmata, including broken blood vessels in nose and larger liver spans. CONCLUSIONS The "best predictor model" showed that older women who were at risk for alcohol misuse or abuse had T-ACE scores of 1 or higher, used two or more OTC drugs regularly, drank large amounts of coffee, used alcohol to fall asleep, and had less sleep latency. Because positive T-ACE scores have high sensitivity and specificity for alcohol abuse, scores of 1 or greater should be addressed in clinical settings, e.g., referrals for more definitive diagnoses and relevant treatment.