The energy cost of walking (at 3.2 km x h(-1)) per unit distance (J x kg(-1) x m(-1)) at gradients of 0%, +7%, and +12% and during a progressive test (2% increase in gradient every 2 min), as well as the overall (aerobic plus anaerobic) net cumulative energy consumption and the corresponding maximal exercise duration were assessed in 19 patients with peripheral arterial disease (PAD) and in 13 moderately active control subjects. With a 0% gradient, the energy cost of walking was approximately 40% greater in patients with PAD than in controls (2.93+/-0.52 and 2.13+/-0.33 J x kg(-1) x m(-1) respectively; P <0.01). In contrast, at gradients of +7% and +12%, the energy cost of walking was similar in the two groups (+7%: PAD, 4.15+/-0.74 J x kg(-1) x m(-1); controls, 4.18+/-0.54 J x kg(-1) x m(-1); +12%: PAD, 5.59+/-1.03 J x kg(-1) x m(-1); controls, 5.64+/-0.75 J x kg(-1) x m(-1)). In patients with PAD, maximal exercise duration with gradients of 0%, +7% and +12% was 449+/-254, 322+/-200 and 229+/-150 s respectively, whereas the net cumulative energy consumption at fatigue was almost constant at approximately 1100 J x kg(-1) for all gradients. The greater energy cost of walking in PAD patients compared with controls in level, but not uphill, walking is interpreted as being mainly the consequence of an altered mechanical locomotory pattern, and not of lower metabolic efficiency. For a wide range of loads, net cumulative energy consumption appears to be independent of maximal exercise duration, a finding that provides a practical criterion for assessing the degree of functional impairment of patients with PAD on metabolic grounds.