The lack of available reference values of arterial PO2, particularly for elderly persons, led us to study a sample of 194 normal nonsmoking subjects, equally distributed over all age ranges from 40 to 90 yr. The radial artery was punctured and blood samples were taken and analyzed on an automated, computerized gas-analyzer. The trend of the mean values of Pa-O2 in the 5-yr class intervals of age showed a clear decline up to the 70- to 74-yr class, and then an inversion, The two regression lines intersecting at this point provided a better fit to the data than did a single regression line (R(2)(2) - R(1)(2) = 0.918 - 0.678 = 0.24; F = 20.49, p = 0.0027). The relationship of Pa-O2 with age was thus subsequently considered for the two subgroups (40 to 74 yr; greater than or equal to 75 yr) identified on the basis of this cutoff. Because of the significant influence on Pa-O2 of age, body-mass index (BMI), and Pa(CO2)in the group 40 to 74 yr of age, the following reference equation was constructed: Pa-O2 (mm Hg) = 143.6 - (0.39 . age) (0.56 . BMI) - (0.57 . PaCO2); R(2) = 0.28; SEE = 7.48; p < 0.0001. For subjects greater than or equal to 75 yr old, for whom there was no correlation with age, BMI, or Pa-CO2, only the mean +/- SD and 5th percentile of Pa-O2 were reported (83.4 +/- 9.15 mm Hg and 68.4 mm Hg, respectively). Pa-CO2 values were not correlated with either age or BMI; the mean +/- SD was 35.79 +/- 3.87 mm Hg. Our study yields appropriate reference values of Pa-O2, including the lower and upper limits for subjects aged 40 to 90 yr, that can be used routinely in pulmonary function laboratories.