Heart rate (HR) is higher during dynamic arm exercise than during leg exercise at equal oxygen consumption levels, but the physiological background for this difference is not completely understood. The vagally mediated beat-to-beat R-R interval fluctuation decreases until the level of approximately 50% of maximal oxygen consumption during an incremental bicycle exercise, but the vagal responses to arm exercise are not well known. Changes in autonomic modulation of HR were compared during arm and leg exercise by measuring beat-to-beat R-R interval variability from a Poincaré plot normalized for the average R-R interval (SD1n), a measure of vagal activity, in 14 healthy male subjects (age 20 +/- 4 years) who performed graded bicycle and arm cranking tests until exhaustion. Seven of the subjects also performed the dynamic arm and leg tests after beta-adrenergic blockade (propranolol 0.2 mg kg-1 i.v.). More rapid reduction occurred in SD1n during the low-intensity level of dynamic arm exercise than during dynamic leg exercise without beta-blockade (e.g. 11 +/- 6 vs. 20 +/- 10 at the oxygen consumption level of 1.2 l min-1; P < 0.001) and with beta-blockade (e.g. 13 +/- 4 vs. 25 +/- 10 at the level of 1.0 l min-1; P < 0.05), and the mean HR was significantly higher during submaximal arm work than during leg work in both cases (e.g. during beta-blockade 81 +/- 12 vs. 74 +/- 6 beats min-1 at the level of 1.0 l min-1; P < 0.05). These data show that dynamic arm exercise results in more rapid withdrawal of vagal outflow than dynamic leg exercise.