Previous research has indicated that the spoken frequency of the self-references "I," "me," and "my" in a structured interview was prospectively related to coronary heart disease (CHD). To assess whether the findings would replicate in another population, we conducted a case-control analysis of 750 structured interviews from the Western Collaborative Group Study. To measure self-references, auditors counted all first person pronouns (I, me, my) and clauses spoken in the audiotaped baseline structured interviews. Matched multiple logistic regression analyses, with or without adjustment for major CHD risk variables, indicated that those who incurred CHD did not self-reference more frequently or densely than the CHD-free control subjects. Type As spoke more clauses and more total self-references but did not have a higher density of self-references than Type Bs. The results question both the method for measuring self-references and the hypothesis that self-referencing are associated with CHD.