Most retrospective cohort studies used the general population as the reference. This entails potential confounding biases related to several lifestyle or constitutional factors. The most important such factor, potentially, for lung cancer studies is smoking. Additionally, retrospective cohort studies of construction workers are limited by some characteristics of this industry: large numbers of relatively small employers, multi-employer work sites, and a highly mobile workforce. Case-control studies, where lifetime work histories can be elicited from study subjects, can overcome some of the limitations of cohort studies, but the comparisons between construction workers and the rest of the population remains vulnerable to confounding. A different paradigm would entail comparing risk of cancer among workers in the construction industry with risks among other blue-collar occupations.