OBJECTIVE: To determine whether high-intensity, progressive gym-based exercise performed once a week is as effective as twice weekly for maintaining both subjective and objective outcomes in older adults post discharge from a metropolitan day rehabilitation centre (DRC). DESIGN: Randomised controlled trial. SETTING: Community-based exercise centre for older adults, located in Metropolitan Adelaide, South Australia. PARTICIPANTS: 21 men and 85 women who completed the DRC programme were assessed and randomly allocated to a study group. INTERVENTION: The two experimental interventions were gym-based exercise programmes (including resistance, aerobic, flexibility and balance training) varying only in frequency of delivery: either once or twice a week, directly compared with usual care (control). MAIN OUTCOME MEASURES: Lower limb strength (one-repetition maximum), balance (Berg Balance Scale), physical function (gait speed, 30-s chair stand test, timed up and go test (primary outcome) and 6-min walk test), self-reported pain (Glasgow Pain Questionnaire), activities of daily living (Barthel Index and Older Americans Resources and Services Multidimensional Functional Assessment Questionnaire), perceived benefits of and barriers to exercise (Exercise Benefits Barriers Scale), quality of life (Assessment of Quality of Life Questionnaire) and exercise frequency preference. RESULTS: Most of the outcomes (69%, 11/16) were maintained over the intervention period with no significant group effects detected between the two intervention groups or compared to the control group. Physical activity levels recorded in the control group showed a significant proportion of participants were actively exercising once weekly. A per-protocol analysis was undertaken to take this potential contamination effect into account. This showed that the control group participants, who did not exercise, did not maintain outcomes to the extent of the intervention groups, with significant group-by-time effects detected between the two intervention groups and the control group. Most of all participants (66%, 62/94) nominated once a week as their preferred exercise frequency. CONCLUSIONS: The overall finding of no significant differences between the two intervention groups for all outcomes measured gives support to the effectiveness of once-a-week exercise in maintaining outcomes at 3 months post rehabilitation. Further research is warranted given the once-a-week exercise intervention should cost less, had higher compliance and was nominated as the preferred exercise frequency by most of the participants.