OBJECTIVES--To examine the effects of drug treatment on laboratory exercise tests in relation to measures of daily activity in patients with chronic heart failure. SETTING--University teaching hospital. SUBJECTS--18 patients with mild to moderate chronic heart failure (New York Heart Association functional class II-III) and 10 age matched healthy controls. METHODS--Assessments were made before and after 12 weeks of vasodilator drug treatment. Exercise capacity was measured during two different types of treadmill exercise, one using a ramp protocol and the other a fixed work load. Corridor walk tests at three self selected speeds were also undertaken and measures of customary activity assessed from pedometer scores. RESULTS--Exercise times were significantly increased from baseline (P < 0.01) with both treadmill protocols after 12 weeks of drug treatment, with a positive correlation between the duration of treadmill exercise for both protocols (r = 0.69, P < 0.01). Corridor walk tests of 100 m at a self selected slow speed also improved (P < 0.02) but these did not correlate with the changes in treadmill exercise time. The pedometer scores of the patients with heart failure were greatly reduced compared with those of the controls (258 (45) x 10(2) v 619 (67) x 10(2) steps/week, P < 0.001) and after 12 weeks of treatment were unchanged (261 (42) x 10(2) steps/week). CONCLUSIONS--These data confirm the need to use different exercise protocols when assessing the benefits of drug treatment in patients with chronic heart failure. Treatments that seem effective with conventional laboratory based exercise tests may not improve daily activities. This may reflect a failure of apparently successful treatment and should be considered when interpreting clinical trials.