IMPORTANCE: Withdrawal of guideline-directed medical therapy (GDMT) for heart failure (HF) is common after atrial fibrillation (AF) catheter ablation and recovery of cardiac function, but safety remains uncertain. OBJECTIVE: To assess the feasibility and safety of phased GDMT withdrawal in patients with AF with highly suspected AF-mediated cardiomyopathy after catheter ablation. DESIGN, SETTING, AND PARTICIPANTS: This open-label pilot randomized clinical trial included adult patients who were enrolled and randomized from April 13, 2023, to September 19, 2024, at Beijing Anzhen Hospital, China, with a 6-month follow-up. Eligible patients were those with suspected AF-mediated cardiomyopathy at 3 months after ablation, defined by sinus rhythm at 3 months after ablation, absence of other suspected cardiomyopathies, improvement of left ventricular ejection fraction (LVEF) from 45% or less to 55% or more, normalized LV end-diastolic diameter (LVEDD), N-terminal pro-brain natriuretic peptide (NT-proBNP) levels less than 250 ng/L, and no HF symptoms or signs. INTERVENTIONS: Patients were randomly assigned in a 1:1 ratio to phased GDMT withdrawal or GDMT continuation. MAIN OUTCOMES AND MEASURES: The primary end point was HF deterioration, defined as LVEF decline more than 10% to less than 55%, LVEDD increase more than 10% and beyond normal, NT-proBNP levels doubling to more than 400 ng/L, or worsening HF signs or symptoms. Secondary outcomes included cardiovascular events, changes in echocardiographic and cardiac magnetic resonance (CMR) parameters, NT-proBNP levels, Kansas City Cardiomyopathy Questionnaire-12 (KCCQ-12) scores, atrial arrhythmia recurrence, and adverse drug events. RESULTS: Among 50 patients enrolled and randomized, 47 completed follow-up (median [IQR] age, 56.0 [48.0-60.5] years; 37 males [78.7%]), including 23 (48.9%) in the GDMT withdrawal group and 24 (51.1%) in the continuation group. HF deterioration occurred in 3 patients with GDMT withdrawal (13.0%) and 0 with GDMT continuation (0%) (P = .11). Early GDMT reinitiation in 3 patients with HF deterioration showed recovery of LVEF or NT-proBNP. No cardiovascular events occurred. Median (IQR) echocardiographic (LVEF: 0% [-3.0% to 3.5%] vs 1.5% [-5.0% to 5.0%]), CMR (LVEF: -2.1% [-7.2% to 4.6%] vs 2.9% [-2.6% to 7.4%]), and KCCQ-12 score (0 [0 to 0.5] vs 0) changes were similar between groups. Median (IQR) NT-proBNP levels declined more in the GDMT continuation group than the withdrawal group (-25.7 [-33.6 to -6.7] pg/mL vs 2.7 [-21.4 to 24.2] pg/mL; P = .03). Adverse drug events were more frequent in the GDMT continuation group than the withdrawal group (5 [20.8%] vs 0 [0%]; P = .050). Arrhythmia recurrence rates were comparable (GDMT withdrawal: 3 [13.0%] vs continuation: 3 [12.5%]). CONCLUSIONS AND RELEVANCE: In this pilot randomized clinical trial of carefully selected patients with AF with normalized cardiac function and sinus rhythm after catheter ablation, 13% of patients with GDMT withdrawal experienced HF deterioration, whereas drug-related complications were more common in the continuation group, suggesting that further studies are needed to determine whether GDMT can be safely discontinued in this population. TRIAL REGISTRATION: Chinese Clinical Trial Registry Identifier: ChiCTR2300077439.