Epidemiology studies of SDB and CVD to date do not provide a conclusive answer to the question of the degree to which SDB impacts CVD or mortality due to CVD. However, most of the studies seem to be consistent with a positive, but perhaps small, association. All the findings discussed or referred to in this review are likely to be biased to some degree. Bias can be both towards underestimation (e.g., from mismeasurment of SDB, and over-control for intermediate factors) and overestimation (e.g., from inadequate control of confounders and improper comparison groups), and the net magnitude of competing biases undoubtedly varies from study to study. Small associations were found in the prospective population-based studies, with one exception. The most obvious methodologic problem in these studies would be likely to result in underestimation of the associations. The case-control studies, in contrast, showed large associations, but serious biases in these studies would probably cause overestimation. Small associations of marginal statistical significance were reported from cross-sectional analyses; findings were limited by sample size. Although each individual study to date could be (and has been ) "dismissed" due to weaknesses, collectively they provide evidence that we cannot dismiss the hypothesis that SDB causes CVD. In many cases, the weak associations can be explained by problems that likely cause underestimation. In fact, finding any association with the limitations of most of the past studies is remarkable. Perhaps most important, the findings to date, in conjunction with biologically plausible mechanisms have sparked the interest needed to initiate the large undertaking of a population-based prospective study. The Sleep Heart Health Study (SHHS)15 is a large multicenter prospective study specifically designed to investigate the role of SDB in incident coronary heart disease, stroke, increased blood pressure, and allcause-mortality. A key feature of the study is that home polysomnography studies are performed on a sample of 6600 men and women, 40 years of age and older, drawn from the samples of other longitudinal studies. The new data collected by SHHS can then be linked to the large amount of data on cardiovascular risk factors available from the "parent" studies. All baseline polysomnography studies have now been completed. Cross-sectional analyses of SDB and CVD history are now being analyzed, and collection of outcome data for longitudinal analyses is underway. Results from SHHS and other studies in the near future should greatly increase our ability to assess the association of SDB and CVD.