Abstract Introduction Randomised controlled trials (RCTs) that fail to meet their recruitment target risk increasing research waste. Acute stroke RCTs experience notable recruitment issues. The efficiency of recruitment to stroke rehabilitation RCTs has not been explored. Aims and objectives To explore recruitment efficiency and the trial features associated with efficient recruitment to stroke rehabilitation RCTs. Methods A systematic review of stroke rehabilitation RCTs published between 2005 and 2015 identified in a search of the Cochrane Stroke Group (CSG) Trials Register from 35 electronic databases (e.g. Medline, CINAHL; EMBASE), clinical trial registers, and hand-searching. Inclusion criteria are stroke rehabilitation intervention, delivered by a member of the rehabilitation team, and clinically relevant environment. We extracted data on recruitment efficiency and trial features. Results We screened 12,939 titles, 1270 abstracts and 788 full texts, before extracting data from 512 included RCTs (n = 28,804 stroke survivor participants). This is the largest systematic review of recruitment to date. A third of stroke survivors screened consented to participate (median 34% (IQR 14–61), on average sites recruited 1.5 participants per site per month (IQR 0.71–3.22), and one in twenty (6% (IQR 0–13) dropped out during the RCT. Almost half (48%) of those screened in the community were recruited compared to hospital settings (27%). Similarly, almost half (47%) those screened at least 6 months after stroke participated, compared to 23% of stroke survivors screened within a month of stroke. When one recruiter screened multiple sites, a median of one stroke survivor was recruited every 2 months compared to more than two per month when there was a dedicated recruiter per site. RCT recruitment was significantly faster per site, with fewer dropouts, for trials conducted in Asia (almost three stroke survivors monthly; 2% dropout) compared to European trials (approximately one stroke survivor monthly; 7% dropout). Conclusions One third of stroke survivors screened were randomised to rehabilitation RCTs at a rate of between one and two per month, per site. One in twenty did not complete the trial. Our findings will inform recruitment plans of future stroke rehabilitation RCTs. Limited reporting of recruitment details restricted the subgroup analysis performed. Trial registration Prospective Register of Systematic Reviews, registration number CRD42016033067.