OBJECTIVE To investigate whether the extent of systolic dysfunction is a useful predictor of the costs of healthcare and social support for patients with heart failure. DESIGN Cross-sectional study with collection of cost data attributed to management of heart failure in the previous year. SETTING Four primary-care practices in Scotland. PATIENTS Patients receiving long term therapy with loop diuretics for suspected heart failure. INTERVENTIONS Two-dimensional and Doppler echocardiography. MAIN OUTCOME MEASURES AND RESULTS Two hypotheses were tested: (i) the proportion of patients incurring costs is higher in patients with abnormal left ventricular (LV) function; and (ii) the median cost per patient that incurs costs is higher in patients with abnormal LV function. Of the 226 patients in the study, 67 (30%) had abnormal systolic function. In comparison with the remaining 159 patients, they had higher healthcare costs [560 Pounds vs 440 Pounds per patient year (1994/1995 values)], were more likely to incur hospital inpatient or outpatient costs [Odds ratio (OR): 2.02; 95% confidence interval (CI): 1.06 to 3.84] and had significantly higher primary-care costs (mean 292 Pounds vs 231 Pounds per patient year; p = 0.02, Mann Whitney test). In contrast, they were no more likely to incur social support costs (OR: 1.22; 95% CI: 0.52 to 2.86) and the mean cost of social support per patient year was lower (234 Pounds vs 373 Pounds). CONCLUSIONS Patients with objectively measured systolic dysfunction incurred significantly higher healthcare costs in the year before diagnosis. This suggests that treatment that improves systolic function will reduce healthcare costs, even in a primary-care population with relatively mild congestive heart failure.