Multiple hospital readmissions for heart failure (HF) are progressively increasing and may be related to continued tobacco and alcohol use. To study this relation, we conducted a retrospective chart audit of all veterans discharged with HF at a large Veterans Administration (VA) facility from 1997 to 1998. Using a multivariate logistic regression model, the smoking and alcohol use of patients who required > 1 HF admission within 1 year were compared with those who did not. Demographic, clinical, and psychosocial variables were also included in the model. Of 753 patients admitted with HF during the review period (mean age 69.1 years, 99% men), 220 patients (29.2%) were readmitted to the hospital at least once (range 1 to 8 readmissions, mean 1.79 +/- 0.27) after the index admission. In a multivariate analysis, current smoking (odds ratio [OR] 1.82; confidence interval [CI] 1.17 to 2.82) and current alcohol use (OR 5.92; CI 3.83 to 9.13) were independent predictors of readmissions. Other predictors included living alone (OR 2.09; CI 1.42 to 3.09), HF associated with ischemic etiology (OR 3.99; CI 2.58 to 6.18), higher New York Heart Association class (OR 2.57; CI 1.86 to 3.55), and care provided by a primary care physician compared with a cardiologist (OR 2.41; CI 1.57 to 3.67). This study confirms that noncompliance to smoking and alcohol restrictions, which are amenable to change, dramatically increases the risk for multiple hospital readmissions among patients with HF. Consequently, evaluation of noncompliance to smoking and alcohol consumption with targeted interventions in this population may be a key component for the reduction of multiple hospital readmissions.