Objectives: To evaluate the use of a combined strategy (a nurse opinion-leader; evidence-based guidelines; a staff education programme and a new recording system) for the implementation of multidisciplinary stroke assessment in an acute hospital setting as measured by the compliance of different professional groups. Design: A quasi-experimental study design utilising a pre-test/post-test group. Setting: Nine medical wards in a 600 bedded outer London Acute NHS Trust (without a stroke unit). Participants: 190 stroke patients (n = 98 pre-test vs n = 92 post-test). Main outcome measures: Professional compliance with assessment guidelines documented in the new recording system. Results: Pre-test compliance with guidelines (n40) ranged from 0% to 100% and post-test ranged from 23-100%. Significant improvements (p < .05) in compliance were found in relation to 25 guidelines and a significant decline was documented in four others. Comparison of pre vs post-test values demonstrated variable changes in compliance with guidelines across professional groups: medical (range -64-+23%); nursing (range +2-+74%); physiotherapy (range -15-+69%); occupational therapy (range 0-+34%). Conclusion: The combined use of an opinion leader, guidelines, education and a recording system had a variable affect on compliance with guidelines; opinion-leadership appeared to have the most influence over compliance. The study design, and guideline specific features may have constrained changes in practice.