Upper-limb aerobic exercise is well tolerated and can improve walking performance in patients with peripheral arterial disease (PAD) (Zwierska et al. 2003). This study aimed to identify physiological adaptations to upper-limb aerobic exercise that could influence the improvement in walking performance. Following approval from the North Sheffield Local Research Ethics Committee, 94 patients (median age 69 years, range 50-85 years) with PAD who were accustomed to the training and assessment protocols, were randomly allocated to arm-crank training (ACT), leg-crank training (LCT) or control groups. Training was performed twice weekly for 24 weeks at equivalent relative exercise intensities (75-85% peak oxygen uptake). Incremental leg-cranking assessments to maximum exercise tolerance were performed before and after the intervention and pulmonary gas exchange variables, blood lactate concentration, ratings of perceived exertion (Borg RPE) and leg pain (Borg CR-10) were recorded. Walking performance, defined as the claudicating distance (CD) and maximum walking distance (MWD) was also assessed using a shuttle-walk protocol (Walker et al. 2000). A mixed-design factorial ANOVA was used to compare groups, with statistical significance set at P < 0.05. Peak oxygen uptake (mean ± S.E.M) for incremental leg-cranking improved after ACT (1.12 ± 0.07 versus 1.33 ± 0.09 l.min-1, P = 0.001) and LCT (1.12 ± 0.05 versus 1.31 ± 0.06 l.min-1, P < 0.001), and was associated with an increased blood lactate concentration in both training groups (P < 0.001), but not in the controls. The CD and MWD improved by 56 ± 9 % and 30 ± 4 % after ACT and by 65 ± 11 % and 35 ± 4 % after LCT, respectively (P at least 0.001). Improvements in MWD were accompanied by comparable increases in leg pain (6 ± 0.5 versus 7 ± 0.5 and 6 ± 0.4 versus 7 ± 0.5), RPE (14 ± 0.6 versus 15 ± 0.6 and 14 ± 0.5 versus 15 ± 0.5) and blood lactate concentration (1.95 ± 0.14 versus 2.40 ± 0.17 and 1.90 ± 0.11 versus 2.26 ± 0.14 mM) after ACT and LCT, respectively (P at least 0.05). The results suggest that ACT evoked similar responses to LCT, and upper-limb aerobic exercise training improved lower-limb exercise capacity via mechanisms that delayed the onset of claudication pain and elevated exercise pain tolerance.