Objective: To investigate the feasibility, acceptability and potential efficacy of group exercise for residents in care homes.Design: Exploratory cluster randomized controlled trial.Setting: Five randomly selected care homes in South Birmingham, UK.Participants: Fifty-six care home residents (mean age 84.5, 71% female), 39 (70%) with cognitive impairments.Intervention: Two homes (n = 28) were randomized to group exercise held twice weekly for five weeks. The remaining three homes (n = 28) formed the control group and received usual care, with no person specifically responsible for exercise training.Outcome measures: Assessments were conducted at zero (pre-intervention), three (post-intervention) and six months (follow-up) using the Rivermead Mobility Index and Hospital Anxiety and Depression Scale or Stroke Aphasic Depression Questionnaire (depending on cognitive impairment). Adherence to group exercise and retention to the study were also documented.Results: No statistically significant improvements in mobility or depression were found in favour of group exercise. Retention to the study was high with 46 (82%) participants completing all assessments. Adherence to group exercise was somewhat lower with participants attending a mean of 3.61 out of 8.5 prescribed sessions (42.5%).Conclusions: Group exercise can be delivered to care home residents with reduced mobility but it is not suitable for residents with severe cognitive impairment. An estimated sample size of 240 participants would be required to detect a clinically significant difference in the Rivermead Mobility Index with 90% power.