Health surveys repeatedly show that females have higher rates of illness, disability days, and health services utilization than do males. Numerous reasons for these sex differences have been hypothesized, based on four general factors thought to differ by sex and to influence self-reported health: genetic characteristics, physical risks, illness behavior, and reporting behavior. This paper briefly reviews these hypotheses, then discusses research strategies to test them. The strategies vary in the type of dependent variables used (medical vs. sociomedical health indicators) and the main method of controlling predictor factors (elimination vs. statistical control). A study was completed in 1978 in the Detroit metropolitan area to (a) explore reasons for sex differences in health and (b) reveal how strongly illness behavior (perception and evaluation of symptoms, propensities to take curative actions or disability days) affects common health indicators, particularly those for the national Health Interview Survey. The study's research design (use of a health diary, and focus on hypotheses about illness behavior) is described.