Objective:To evaluate the feasibility of a multicentre, observer-blind, pilot randomized controlled trial (RCT) of a wristband accelerometer with activity-dependent vibration alerts to prompt impaired arm use after stroke.Design:Parallel-group pilot RCT.Setting:Four English stroke services.Participants:Patients 0–3 months post stroke with a new arm deficit.Intervention:Participants were randomized to wear a prompting or ‘sham’ wristband during a four-week self-directed therapy programme with twice-weekly therapy review.Main outcomes:Recruitment, retention and adherence rates, safety and completion of assessments were reported. Arm recovery was measured by Action Research Arm Test (ARAT) and Motor Activity Log (MAL) without statistical comparison.Results:In total, 33 patients were recruited (0.6 per month/site; median time post stroke: 26 days (interquartile range (IQR):15.5–45)). Baseline, four-week and eight-week median (IQR) ARAT for the control group (n = 19) were 15 (2–35), 35 (15–26) and 31 (21–55) and those for the intervention group (n = 14) were 37 (16–45), 57 (29–57) and 57 (37–57), respectively; for MAL Amount of Use, the corresponding values in the control group were 0.2 (0.0–1.2), 1.1 (0.3–2.9) and 1.2 (0.7–2.9) and in the intervention group were 1.4 (0.5–2.6), 3.8 (1.9–4.5) and 3.7 (2.1–4.3). Four participants withdrew from the study. Wristbands were worn for 79% of the recommended time. The intervention and control group participants received a median of 6.0 (IQR: 4.3–8.0) and 7.5 (IQR: 6.8–8.0) therapy reviews. A median of 8 (IQR: 6–10) prompts were delivered per intervention participant/day. Research assessments were completed for 28/29 and 25/28 patients at four and eight weeks. Eight serious adverse events were reported, all unrelated to the intervention.Conclusion:A multicentre RCT of wristband accelerometers to prompt arm activity early after stroke is feasible. A total sample of 108 participants would be required.