The U.S. Environmental Protection Agency has sponsored a field study of personal exposures to carbon monoxide (CO) in Washington, D.C., and Denver. CO concentrations in exhaled breath were measured in 837 nonsmoking adult residents of Washington and 454 nonsmoking adult residents of Denver between November 8, 1982, and February 25, 1983. The percent of people exceeding 10 ppm (about 2% COHb saturation) was 10% in Denver and 5% in Washington. Mean (+ or - S.E.) breath concentrations were 7.2 + or - 0.2 ppm in Denver and 5.1 + or - 0.2 ppm in Washington. These concentrations agreed well with a previous nationwide study of blood carboxyhemoglobin (NHANESII), but were higher than would be predicted from outdoor fixed stations. Comparisons of COHb estimated from breath measurements and COHb estimated by CO exposures and the Coburn-Forster-Kane (CFK) equation indicate a discrepancy, with the breath COHb tending towards higher estimates. Possible contributing factors to the discrepancy are uncertainties in published breath CO-blood COHb correlations, a tendency for the personal monitors to drift with time, and uncertainties in physiological function values used in the CFK equation.