The findings of many studies conducted before 1978 suggest that Type A behavior (TAB) contributes to the development of coronary heart disease (CHD). In contrast, many recent studies have found no association between these variables. Through meta-analysis, several reasons for null findings are identified. First, a type of range restriction bias, disease-based spectrum (DBS) bias, produced many null findings. A study is vulnerable to DBS bias when researchers select only high-risk or diseased Ss for study. Second, self-report measures of TAB were often associated with null findings. Finally, null results were found for all studies that used fatal myocardial infarction as a disease criterion. In addition to identifying the reasons for null findings, this research suggests that TAB, as assessed by the structured interview, is associated with CHD. More Type As (70%) were found in diseased populations of middle-aged men than in healthy populations of middle-aged men (46%).