Coronary heart disease incidence during 1958-1974, in a cohort of Hiroshima and Nagasaki residents, was found to relate significantly (p = 0.01) to total leukocyte count, taken an average of two years earlier. Relative risks, as a function of leukocyte count, did not appear to depend on sex or cigarette smoking status, but may be larger for subjects less than 65 years of age than for older persons. This study examines, for the first time, differential leukocyte counts and percentages in relation to coronary heart disease incidence. Both neutrophil count and eosinophil count significantly relate to coronary heart disease incidence, while there is also a suggestion for a relationship with monocyte count. A possible immunopathologic basis for such associations is mentioned. It will be important for other studies to confirm these relationships, however, as the present data do not allow one to clearly show the relative risk, at a specified total leukocyte count, to depend on the differential leukocyte fractions. Available data also leave uncertainty concerning the "independence" of leukocyte counts in the specification of coronary heart disease risk, relative to such known risk factors as cigarette smoking habits and serum cholesterol levels.