AIMS: Reduced skeletal muscle mass is a major component of sarcopenia, associated with impaired exercise capacity and poor prognosis in patients with heart failure (HF). Measurement of skeletal muscle mass by dual-energy X-ray absorptiometry may be affected by fluid retention, typically in the patients' lower extremities. The aim of the present study was to elucidate the association between upper and lower extremity skeletal muscle mass (USM and LSM) and all-cause mortality in hospitalized patients with HF, after discharge. METHODS: This was a single-centre observational cohort study of 418 patients (59% were men) admitted with a diagnosis of HF (71 ± 13 years), with a left ventricular ejection fraction of 39 ± 16%. USM and LSM were measured by dual-energy X-ray absorptiometry with patients in a stable state after decongestion therapy. RESULTS: The USM and LSM were 5.29 ± 1.18 and 13.78 ± 3.20 kg for men and 3.37 ± 0.68 and 9.19 ± 1.80 kg for women. A positive correlation was obtained between USM and LSM with mid-upper arm circumference (r = 0.684, P < 0.001) and calf circumference (r = 0.822, P < 0.001), respectively. During a median follow-up of 37 months, 92 (22.0%) of the 418 patients died. A Kaplan-Meier analysis revealed that sex-specific quartiles of USM/height2 and LSM/height2 were associated with all-cause mortality (both P < 0.001 by the log-rank test). In Cox models adjusted by age, sex, creatinine, haemoglobin, NYHA class, and height2 , the hazard ratio with 95% confidence intervals for all-cause mortality was 0.557 [0.393-0.783] (P < 0.001) for USM per 1 kg, and 0.783 [0.689-0.891] (P < 0.001) for LSM per 1 kg. The receiver-operator-characteristic curve analysis showed a comparable area under the curve between the USM/height2 and LSM/height2 (0.557 vs. 0.568, P = 0.562) in predicting all-cause mortality. The ratio of USM to LSM was significantly lower in 37 patients with residual leg oedema than in the 360 patients without oedema (36.1% vs. 38.1%, P = 0.004), suggesting the influence of oedema on measured LSM. CONCLUSIONS: Both USM and LSM had a prognostic implication on mortality after discharge in HF, even though LSM may have been affected by leg oedema. These findings indicate that clinicians should not ignore a patient's USM or LSM in the prognostication of patients with HF.