A rapid (within 0-5 s) increase in skeletal muscle blood flow has been demonstrated following muscle contraction, yet the mechanism remains unresolved. Recently, it was suggested that the entire rapid exercise hyperemia could be attributed to the mechanical muscle pump effect. Other evidence indicates that the muscle pump cannot increase arterial flow. We measured human forearm blood flow with the arm positioned above or below heart level during 1) simulation of rhythmic muscle pump function via repeated inflation/deflation of a forearm cuff to 100 mmHg to achieve mechanical emptying of forearm veins, and 2) 1-s single-cuff inflations, 1-s voluntary forearm contractions, and 1-s contractions performed within a cuff inflation. Rhythmic cuff inflation increased blood flow with the arm below heart level (P < 0.05) but not above. Flow following single contractions was higher than flow following cuff inflation within 2 s (P < 0.05). Peak flow increases due to a single mechanical venous emptying (7.7 +/- 0.7 ml.100 ml(-1) min(-1)) could account for 60% of the peak flow increase due to muscle contraction (12.8 +/- 1.0 ml.100 ml(-1).min(-1)) with the arm below heart level, whereas above heart level mechanical venous emptying accounted for 46% of the flow increase due to contraction (3.0 +/- 0.4 vs. 6.5 +/- 0.6 ml.100 ml(-1).min(-1)). We conclude that a functional muscle pump does exist in the human forearm in vivo, but that a rapid vasodilation detectable within 2 s also contributes to the early exercise hyperemia.