The time course of the pulmonary vascular response to hypoxia in humans has not been fully defined. In this investigation, study A was designed to assess the form of the increase in pulmonary vascular tone at the onset of hypoxia and to determine whether a steady plateau ensues over the following approximately 20 min. Twelve volunteers were exposed twice to 5 min of isocapnic euoxia (end-tidal Po(2) = 100 Torr), 25 min of isocapnic hypoxia (end-tidal Po(2) = 50 Torr), and finally 5 min of isocapnic euoxia. Study B was designed to look for the onset of a slower pulmonary vascular response, and, if possible, to determine a latency for this process. Seven volunteers were exposed to 5 min of isocapnic euoxia, 105 min of isocapnic hypoxia, and finally 10 min of isocapnic euoxia. For both studies, control protocols consisting of isocapnic euoxia were undertaken. Doppler echocardiography was used to measure cardiac output and the maximum tricuspid pressure gradient during systole, and estimates of pulmonary vascular resistance were calculated. For study A, the initial response was well described by a monoexponential process with a time constant of 2.4 +/- 0.7 min (mean +/- SE). After this, there was a plateau phase lasting at least 20 min. In study B, a second slower phase was identified, with vascular tone beginning to rise again after a latency of 43 +/- 5 min. These findings demonstrate the presence of two distinct phases of hypoxic pulmonary vasoconstriction, which may result from two distinct underlying processes.