Long-term heart rate (HR) control is a management strategy for patients with chronic atrial fibrillation (AF). Nevertheless, the optimal target HR of AF patients is debatable. Our aim was to study HR at rest, during, and after a 6-minute walk test (6MWT) in AF patients, compared with controls with sinus rhythm (SR). Consecutive matched patients with AF (n = 186) or SR (n = 172) were recruited, and 6MWT was performed. HRs at rest, during 6MWT, and recovery periods were recorded. All subjects were divided into 5 subgroups (<80 beats/min, 80 to 89 beats/min, 90 to 99 beats/min, 100 to 109 beats/min, and ≥110 beats/min) according to the HR at rest. No statistical difference was observed in baseline HR at rest, between AF and SR groups (p = 0.30). The exercise HR increase percentage was significantly higher in overall AF patients compared with those in SR (40 ± 15% vs 14 ± 7%, p <0.001). Even with similar mean baseline HRs at rest, the 5 AF subgroups all showed significantly higher mean exercise HR, maximal exercise HR, and maximal exercise HR increase percentage compared with their respective SR subgroups, especially the subgroups with HR at rest >90 beats/min. Unlike the SR patients, the 4 AF subgroups with HR >80 beats/min at the fifth minute after 6MWT did not recover to at rest levels. In conclusion, HR increased excessively during 6MWT and HR recovery was delayed after 6MWT in AF patients, especially when HR at rest is >90 beats/min. The optimal initial HR at rest for AF patients should perhaps be <90 beats/min.