Cardiovascular disease is the leading cause of mortality in patients whose lives depend on hemodialysis. We developed a method for measuring cardiac output (CO) and central blood volume (CBV) in hemodialyzed patients that may help to elucidate the mechanisms and consequences of cardiac disease in this population. This report describes the technique, focusing on the main sources of error and how they can be prevented. Three principal sources of error were identified: (1) access recirculation (existing or induced during injection); (2) the second pass of the indicator through the cardiopulmonary system, exacerbated by prolonging the duration of intravenous injection; and (3) the transit time of the indicator through the dialysis blood lines. After the algorithms were adjusted to prevent the above errors, the reproducibility of CO and CBV, expressed as the absolute percent deviation from the average of duplicates (3,488 values duplicated within 5 minutes), was 4.3 +/- 3.8% for CO and 4.1 +/- 3.8% for CBV. To determine the clinical value of routine CO and CBV measurements, morbid events (nausea, vomiting, and/or muscle cramps) were prospectively recorded in 73 randomly selected hemodialysis patients. CO and CBV were measured near the beginning and near the end of 98 dialysis sessions during which 28 morbid events were identified. In 10 of these sessions, where morbid events took place within 30 minutes of the measurements, CBV appeared to be a more sensitive indicator of morbid events than CO. We conclude that CO and CBV can be routinely and reliably measured during hemodialysis if precautions are taken to avoid specifically identified sources of error. Preliminary studies suggest that these measurements may have significant prognostic value.