The influence of the xanthine derivative pentoxifylline ('Trental' or BL191; Hoechst-Roussel) on exercise tolerance was measured in 38 subjects with stable, severe to moderately severe, intermittent claudication who completed a randomised, double-blind, placebo controlled, cross-over clinical trial. Patients received placebo tablets or 400 mg slow-release pentoxifylline tablets ('Trental 400') twice a day for one week, followed by three times daily for seven weeks, and then crossed over to receive the alternate preparation for another eight weeks. Claudication distance and walking distance were measured on a treadmill before starting treatment and again at four-week intervals during the trial. At the same times, red blood cell filterability, plasma fibrinogen concentration and blood viscosity, resting and post-ischemic calf muscle blood flow, and the resting and post-exercise ankle/brachial systolic pressure ratio were also measured. In this study, the observed effects of pentoxifylline treatment were no greater than those of placebo, even though serum levels of pentoxifylline and its hydroxy-metabolite were within the anticipated range. This was shown by a 'therapeutic effect ratio' of 0.98 for treadmill claudication distance and 0.96 for treadmill walking distance after within-patient analysis at the end of the cross-over (where a ratio of 1.0 means the test drug and placebo effects are identical). These ratios have 95% confidence limits of 0.72-1.34 and 0.74-1.25, respectively.