Approximately two-thirds of patients implanted with ventricular assist devices recover sufficiently to requalify for heart transplantation, and the other one-third die of complications that are often secondary to delayed ventricular assist device implantation and subsequent end-organ failure. To determine whether any preoperative predictors of survival exist, univariate statistics and multivariate stepwise logistic regression analysis were performed on pre-ventricular assist device demographics, hemodynamics, and blood chemistry in 186 patients receiving Thoratec ventricular assist devices (Thoratec Laboratories Corp., Berkeley, Calif.) while awaiting transplantation. The duration of circulatory support averaged 19.6 days (maximum, 226 days). One hundred thirty-seven patients (74%) received biventricular support, 47 received isolated left ventricular assist devices, and two received right ventricular assist devices. The average blood flow was 5.0 +/- 0.9 L/min. One hundred eighteen patients (63%) ultimately received heart transplants, of whom 96 patients were discharged. Age, gender, weight, and diagnosis were not related to survival, nor were preoperative cardiac index, pulmonary capillary wedge pressure, intraaortic balloon pumps, or cardiac arrests. Pre-ventricular assist device creatinine levels (p = 0.24) and total bilirubin levels (p = 0.09) were not significant, but blood urea nitrogen level (p = 0.02) and previous operations (p = 0.05) were related to survival, using univariate techniques. Patients with cardiac operations more than 30 days previously had the lowest survival-to-transplantation (39%) compared with patients with no previous operations (67%) or operations within the previous 30 days (61%). Blood urea nitrogen level was the only parameter found to be significant (p = 0.016) in a multivariate model.(ABSTRACT TRUNCATED AT 250 WORDS)