Background: It is unclear how arm use in daily life changes after stroke since studies investigating the change in arm use poststroke are scarce. The aim of this study was to investigate the change in arm use during the first six months poststroke. Secondary aim was to compare arm use changes between arm recovery clusters. Methods: Arm use was measured during week 3, 12, and 26 poststroke with accelerometers on the wrists and the nonaffected leg. Outcomes were the amount of affected and nonaffected arm use during sitting and standing per day and per sit/stand hour, and the daily ratio between arms. Arm function was measured with the Fugl-Meyer Upper Extremity Scale to identify recovery clusters (poor/moderate/excellent). Generalized estimating equations compared arm use outcomes between time points and between recovery clusters. Results: Thirty-three stroke patients participated. Affected arm use per day increased between week 3 and 12 (30 %; p = 0.04) and it increased per sit/stand hour between week 3–12 (31 %; p < 0.001) and between week 3 and 26 (48 %; p = 0.02). Nonaffected arm use per day decreased between week 3 and 12 (13 %; p < 0.001) and between week 3 and 26 (22 %; p < 0.001) and it decreased per sit/stand hour between week 3 and 26 (18 %; p = 0.003). The daily ratio increased between week 3 and 12 (43 %; p < 0.001) and between week 3 and 26 (95 %; p < 0.001). Changes in arm use did not differ significantly between recovery clusters (p = 0.11–0.62). Affected arm use was higher in the excellent recovery cluster (p < 0.001). Conclusions: Affected arm use and the ratio between arms increase during the first 26 weeks poststroke especially in patients with excellent arm recovery.