Abstract The objective of this study was to describe the predictive value of dyspnea at rest in the diagnosis of heart failure in older adults. We have conducted a retrospective chart review of older adults hospitalized with heart failure in 11 hospitals in Alabama, US. We defined dyspnea at rest as dyspnea not accompanied by activities, sleep, or lying down. We confirmed diagnosis of heart failure using modified Framingham criteria (based on available variables). We estimated values for sensitivity, specificity, predictive value positive and likelihood ratio positive of dyspnea at rest along with their 95% confidence intervals (95% CI). Patients had a mean (±S.D.) age 79 (±7.5) years, were 60% female and 18% African–American. The prevalence (pre-test probability) of heart failure was 77%. Dyspnea at rest was reported by 975 (89%) patients. Dyspnea on exertion, orthopnea and paroxysmal nocturnal dyspnea were, respectively, reported by 279 (26%), 356 (33%) and 230 (21%) patients. The sensitivity, specificity and positive predictive value (post-test probability) of dyspnea at rest were, respectively, 92% (95% CI=90–94%), 19% (95% CI=14–24%) and 79% (95% CI=77–82%). Patients with dyspnea at rest were 13% (likelihood ratio=1.13; 95% CI=1.06–1.20) more likely to have heart failure than those without. Presence of any three of the symptoms of dyspnea at rest, dyspnea on exertion, orthopnea, paroxysmal nocturnal dyspnea, fatigue and lower extremity edema increased the positive predictive value to 90% (95% CI=87–93%). Dyspnea at rest was the commonest symptom among elderly hospitalized HF patients and had high sensitivity, but low specificity and positive predictive value, which increased in the presence of other symptoms.