In six normal upright subjects, a 100 mol bolus-composed of equal parts of neon, carbon monoxide, and acetylene (Ne, CO, and C(2)H(2))-was inspired from either residual volume (RV) or functional residual capacity (FRC) during a slow inspiration from RV to total lung capacity (TLC). After breath holding and subsequent collection of the exhalate, diffusing capacity and pulmonary capillary blood flow per liter of lung volume (D(L)/V(A) and Q(C)/V(A)) were calculated from the rates of CO and C(2)H(2) disappearances relative to Ne. The means: D(L)/V(A) = 5.26 ml/min x mm Hg per liter (bolus at RV), 6.54 ml/min x mm Hg per liter (at FRC); Q(C)/V(A) 0.537 liters/minute per liter (bolus at RV), 0.992 liters/minute per liter (at FRC). Similar maneuvers using Xenon-133 confirmed that, during inspiration, more of the bolus goes to the upper zone if introduced at RV and more to the lower, if at FRC. A lung model has been constructed which describes how D(L)/V(A) and Q(C)/V(A) must be distributed to satisfy the experimental data. According to this model, there is a steep gradient of Q(C)/V(A), increasing from apex to base, similar to that previously determined by other techniques-and also a gradient in the same direction, although not as steep, for D(L)/V(A). This more uniform distribution of D(L)/V(A) compared with Q(C)/V(A) indicates a vertical unevenness of diffusing capacity with respect to blood flow (D(L)/Q(C)). However, the relative degree of vertical unevenness of D(L)/V(A) compared with Q(C)/V(A) can account only in part for previous observations attributed to the inhomogeneity of D(L)/V(A) and Q(C)/V(A). Thus, a more generalized unevennes of these ratios must exist throughout the lung, independent of gravitation.