The aim of this study was to evaluate the construct validity of the SAIB in a community sample of elderly people. The SAIB was administered to a large community sample representative of the German population aged 60-85 years (n=1593). The original model was assessed and then refined through confirmatory and exploratory factor analyses. Criterion validity was evaluated by comparing SAIB scores with external criteria in 3 categories: subjective health, chronic illness and health care utilization. The originally suggested five factor structure of the SAIB yielded a comparative fit index (CFI) of 0.70 and the weighted root mean square residual (WRMR) was 3.68. A shortened questionnaire with 13 items and four factors resulted in better model fit (CFI 0.97 and WRMR 1.3). Correlations between subjective health and the new scales ranged from 0.06 to 0.33. Effect sizes (Cohens d) of mean differences in factor scores between those with and without healthcare system contact varied by healthcare type, ranging from 0.05 to 0.94; effect sizes were largest in relation to contact with psychotherapy and alternative medicine practitioners. We propose a shortened version of the SAIB with a different scale structure, which resulted in better model fit with our data. Neither the original nor revised SAIB appeared to discriminate well in terms of health care use, suggesting that the illness behavior as currently conceptualized may not fully explain the increased use of healthcare in the elderly.