Objective: To examine the effectiveness of a set of rules for referral and therapy input in a three-tiered physiotherapy program on activities of daily living (ADL), motor function, and quality of life of stroke survivors. Design: Randomized controlled study. Setting: Rehabilitation departments of 11 teaching hospitals. Subjects: A total of 285 participants with stroke. Outcome measures: Primary outcome was ADL independence measured with the Modified Barthel Index (MBI) at weeks 3, 6, 9, 13, and 17. Secondary outcomes were motor function and quality of life measured with Fugel-Meyer Assessment (FMA) and Stroke-Specific Quality-of-Life (SSQOL) scale. Intervention: Two complementary sets of rules governing rehabilitation delivery were introduced: a set of criteria that determined when someone ought to move from tier 1 onto tier 2, and from tier 2 onto tier 3, and a second set of rules that determined the amount and type of physiotherapy input given in each tier. Control group participants received conventional rehabilitation without any specified guidelines. Results: With a difference of 3.97 (95% confidence interval (CI): 1.59-6.36), MBI increased stronger in the study group than in controls between baseline and week 3 (P = 0.001). This difference could be sustained until study end-point. No significant differences were found for FMA. Differences in increase of SSQOL were higher in the intervention than control at week 9 (P < 0.05). Conclusion: Introduction of a set of rules for referral and therapy input at different stages of rehabilitation partially improved patients' ADL and quality of life, but did not improve motor function.