We thank Dr Hiatt for his remarks on our meta-analysis.1 Our conclusions on physical exercise and smoking cessation had to be weakly worded since data from adequately controlled trials are scarce. Thirty-three uncontrolled studies were excluded from our review and only 6 relatively small, open, randomized trials were identified. As reported in their Table 1,1 a supervised exercise program was prescribed in all randomized trials, with the exception for Larsen and Lassen's2 study. In the trial by Larsen and Lassen, the improvement in walking distance resulted significantly in favor of treated patients while in another study by Lundgren et al3 with more patients trained in an exercise program under supervision, it was not. Consequently, it is difficult to make a case for supervised training based on the data from these trials alone. The important issue of comparing supervised with unsupervised programs has been addressed by 2 recent small studies4,5 (Table 1. Patterson et al,4 in a small trial, demonstrated a superiority of supervised over home-based treatment in improving walking time, while quality of life was equally affected in the 2 groups. Regensteiner et al5 showed a significant improvement both in peak walking time and functional status in the supervised group. However, a calculation of differences between functional status scores obtained at the end of treatment shows no significant differences among groups. These 2 small studies lend some support to the superiority of the supervised programs over the general advice to walk and exercise more. In contrast, Gardner and Poehlman6 demonstrated that among the different components of a training program, the degree of supervision was not relevant (P = .476) in the improvement of pain-free walking. Furthermore, the mode of exercise program associated with a significant gain in pain-free walking was walking only (P = .042), compared with the combination of other exercises. Others7 disagree with these observations and, based on their critical analysis of randomized trials (small sample sizes, absence of blind assessment of the outcome, short-term results), clearly show the urgent need for the definition of the optimal and most cost-effective exercise program for these patients. In addition, patient compliance seems essential, depending on "belief in training and motivation to participate."8 Instead of including patients in expensive hospital-based programs, it possibly may be enough to strongly motivate and encourage them to exercise (in groups) outside the hospital (eg, at home or the gymnasium). The need for large randomized studies addressing these issues is ample.