This study evaluated impedance cardiography (ZCG) estimates of stroke volume (SV) during exercise. Seven subjects were studied at rest and during progressive cycle exercise in supine and upright positions. SV was determined by ZCG (SVZCG) during exercise and for the first 5 cardiac cycles following exercise. SVZCG was compared with separate measurements of SV by CO2 rebreathing (SVCO2). Static blood resistivity (p) was measured at each level of exercise. No significant differences were found between supine exercise and immediate post-exercise values for the peak of the first derivative of the impedance change (dZ/dtmax), left ventricular ejection time (LVET), or SVZCG. Small differences in dZ/dtmax and SVZCG, but not LVET, were found in exercise to post-exercise cycling in the upright position. Intra-individual SVZCG and SVCO2 were moderately correlated (upright mean r = 0.64, supine r = 0.42) from rest to 70% of peak VO2. Similar correlations were found between Pulse-O2 (VO2/heart rate, used as an index to SV) and both SVZCG (upright r = 0.73, supine r = 0.57) and SVCO2 (upright r = 0.8, supine r = 0.65). The ZCG parameters dZ/dtmax and LVET correlated better with Pulse-O2 (dZ/dtmax: upright r = 0.92, supine r = 0.73; LVET: upright r = -0.9, supine r = -0.9). SVZCG calculated with the Kubicek equation performed as well as SVCO2. ZCG might be a superior method if the inversely correlated parameters, dZ/dtmax and LVET, were not expressed as a product to calculate SV.