It has been suggested that Coxsackie B virus infections may play a part in causing or triggering myocardial infarction. This study was designed to compare the incidence of such infections in Coronary Care Unit patients and normal controls. The choice of a suitable criterion for diagnosis of Coxsackie infection is discussed fully. Two hundred and fifty admissions to a Coronary Care Unit and 100 control subjects had a serum sample tested by microneutralisation for Coxsackie B antibodies. The incidence of infection among 130 patients diagnosed as acute myocardial infarction was 5% compared with 4% in the control group. In a subgroup classified as non-transmural myocardial infarction, the incidence of infection was 14%. The sex ratio of this group differed from the myocardial infarction group as a whole suggesting that the non-transmural group may not have been homogeneous. Normal coronary arteriograms were subsequently found in three patients who were diagnosed as non-transmural myocardial infarction but who had serological evidence of recent Coxsackie infection. This study does not demonstrate an association between Coxsackie infection and myocardial infarction as a whole and does not support the view that Coxsackie infection causes or provokes myocardial infarction. It does, however, suggest that myocarditis may simulate non-transmural infarction.