Heart failure (HF) in the community differs meaningfully from that in clinical trials, particularly the higher prevalence of patients with preserved left ventricular (LV) ejection fraction (EF) typically excluded from clinical trials, thus limiting knowledge of their responsiveness to beta-blocker therapy. From a community-based registry of 4,280 patients with HF starting treatment with the beta blocker carvedilol, we compared characteristics, carvedilol titration, and outcomes of patients according to LVEF >40% or <40% (as in clinical trials) and across the spectrum of LVEF <21%, 21% to 30%, 31% to 40%, and >40%. Patients with preserved EF (LVEF >40%) were older and more often women and hypertensive. Lower LVEF was associated with worse functional class and more HF hospitalizations in the previous year. Carvedilol dose decreased with increasing LVEF. Hospitalization rates for HF related inversely to LVEF before starting carvedilol therapy and decreased from the previous year in all LVEF groups during follow-up. Although 1-year mortality rate decreased from 8% with LVEF < or =20% to 6% with LVEF >40%, adjusted hazard ratios were not significantly different across LVEF groups. Thus, characteristics of community patients with HF vary across the spectrum of LVEF. Patients with HF and preserved EF treated with carvedilol in the community improve symptomatically and experience fewer HF hospitalizations after initiating carvedilol. In conclusion, without a control group, the effect of carvedilol on outcomes is not conclusive and trials of carvedilol in patients with HF and preserved EF should be undertaken.