Background Hemodynamics often are used as surrogate end points in phase II trials of acute heart failure (HF). We reviewed the Flolan International Randomized Survival Trial (FIRST) database to identify the hemodynamic variables that best predict survival in patients with advanced HF receiving epoprostenol therapy and to determine whether hemodynamics could predict the overall effect of a drug. Methods The trial enrolled 471 patients with class IIIb/IV HF and ejection fraction ≤25% for ≥3 months, all of whom underwent screening pulmonary artery catheter insertion. Patients were randomly assigned to receive either epoprostenol (n = 201) or placebo (n = 235); epoprostenol therapy was guided by pulmonary artery catheter measures, and standard treatment was guided by clinical findings. Multivariable modeling was used to identify and quantify the demographic, clinical, and hemodynamic variables most associated with 1-year survival. Results In multivariable modeling, HF class, decreased pulmonary capillary wedge pressure (PCWP), and age best predicted 1-year survival. After adjustment for age and HF class, decreased PCWP still significantly predicted survival (hazard ratio, 0.96 for every 1–mm Hg decrease; 95% confidence interval, 0.94 to 0.99; P = .003). Survival was significantly higher with decreases in PCWP ≥9 versus <9 mm Hg, even after adjustment for age and HF class. Survival of patients in the PCWP ≥9 group was comparable with, and that of the PCWP <9 group was significantly higher than, survival of patients in the control group (hazard ratio, 1.44; 95% confidence interval, 1.05 to 1.99; P = .024). Conclusions The reduction in PCWP was the hemodynamic measure most predictive of survival in patients with advanced HF. However, patients with a ≥9–mm Hg decrease had no better survival than patients in the control group, who had limited changes in hemodynamics. Thus, improvement in hemodynamics may not predict the overall effect of a drug. (Am Heart J 2001;141:908-14.)