This work aimed to study the efficacy and safety of the class III antiarrhythmic agent cavutilide (Niferidil, Refralon) for pharmacological cardioversion in patients with paroxysmal and persistent atrial fibrillation (AF) and heart failure (HF). Methods and Results: In this retrospective cohort study, 58 patients with stable HF (aged 69 [61;73] years, 30 males, 78% with persistent AF) and 274 patients without HF (aged 63 [57;70] years, 196 males, 56% with persistent AF) were included. The median AF duration in the group with HF was 35.5 [10.6;124] days, and that in the group without HF was 14.5 [3.6;90] days. All patients received 5–30 µg/kg cavutilide intravenously in one to four (if needed) boluses of 5–5–10–10 µg/kg at 15 min intervals. Subsequent boluses were not administered if the patient’s sinus rhythm (SR) was restored or if bradycardia, QT prolongation > 500 ms or evidence of proarrhythmia was observed. Holter electrocardiogram monitoring was started before infusion and was continued for 24 h. The main criterion for an antiarrhythmic effect was sinus rhythm restoration within 24 h of the initial bolus. Results: Cavutilide converted AF to SR in 37.9% of patients with HF after bolus 1 (5 µg/kg), in 58.6% after bolus 2 (cumulative dose = 10 µg/kg), in 74% of cases after bolus 3 (cumulative dose = 20 µg/kg) and in 92.8% of cases after bolus 4 (cumulative dose = 30 µg/kg). Cavutilide was effective in 89% of cases with persistent AF with a median duration of 70.5 [30;159] days and in 92% of cases with paroxysmal AF with a median duration of 36 [24;102] h. In the group of patients without HF, the effectiveness of bolus 1 was 36.9%, that of the bolus 2 was 58%, that of the bolus 3 was 77% and that of the bolus 4 was 90.1%. Cavutilide restored SR in 90% of patients with persistent AF with a median duration of 82.5 [28;180] days and in 90% of cases with paroxysmal AF with a median duration of 50 [24;120] h. No statistically significant difference in the probability of SR restoration or the effectiveness of each bolus of cavutilide was found between patients with and without HF. The median time to restoration of SR in patients with HF was 23 [11;55] min, and that in patients without HF was 22 [10;45] min (p = 0.424). No cases of symptomatic/severe bradycardia were observed in either group. QT prolongation over 500 ms after cavutilide injection was registered in 19% of patients without HF and in 15.5% of those with HF (p = 0.58). Short runs of Torsade de pointes tachycardia occurred in one patient (0.4%) without HF after 10 µg cavutilide administration and were successfully treated with MgSO4. Conclusions: Cavutilide demonstrated a high likelihood of AF conversion to SR in paroxysmal (92%) and persistent (89%) arrhythmia and HF. Concomitant HF and its severity do not affect the efficacy and safety of cavutilide.