Four hundred seventy-eight male employees of the Iowa Department of Public Safety were questioned regarding usage of cigarettes and coffee, prior to a 52 week prescribed exercise training program. Subjects were classified as smokers and nonsmokers across five coffee usage categories, ranging from nondrinkers to heavy drinkers (2 x 5 factorial). Measurements taken during entry into the program allowed analysis across nine dependent variables, believed to be linked to coronary heart disease. Dependent effects included; (a) FEV(,1), (b) total serum cholesterol, (c) triglycerides, (d) HDL-c, (e) resting systolic blood pressure, (f) resting diastolic blood pressure, (g) resting heart rate, (h)(' )VO(,2)max, and (i) max double product. Variability due to age, weight, activity level, and body composition was removed prior to analysis. Multivariate analysis of variance indicated no overall coffee by cigarette usage interaction and no overall main effect of coffee consumption. An overall main effect of smoking was detected, with follow-up analysis revealing higher levels of HDL-c in nonsmokers, lower systolic and diastolic blood pressures in smokers, and higher(' )VO(,2)max values for nonsmokers. Smokers also had lower FEV(,1) percentages, but only at the older ages. Although statistically significant, the adjusted mean HDL-c values differed by only 4 mg/dl, systolic and diastolic blood pressure by 2-3 mg, Hg, and VO(,2)max by 1 liter/min. The amount of explained variance by whether a person smoked or not was less than 3 percent for HDL-c and systolic blood pressure and less than 1 percent for diastolic blood pressure and VO(,2)max. Although it was concluded that these differences are probably real, they are unlikely to be important from a physiological and clinical standpoint.