Background: Randomised controlled trials demonstrating improved longevity are needed to justify high-dose vitamin D supplementation for older populations. Objectives: To demonstrate the feasibility of a large trial (n ≈ 20,000) of high-dose vitamin D in people aged 65–84 years through general practitioner (GP) practices, and to cluster randomise participating practices between open-label and double-blind randomisation to compare effects on recruitment, compliance and contamination. Design: Twenty GP practices were randomised in matched pairs between open-label and double-blind allocation. Within each practice, patients were individually randomised to vitamin D or control (i.e. no treatment or placebo). Participants were invited to attend their GP practice to provide a blood sample and complete a lifestyle questionnaire at recruitment and again at 2 years. Randomisation by telephone followed receipt of a serum corrected calcium assay confirming eligibility (< 2.65 nmol/l). Treatment compliance was reported by quarterly follow-up forms sent and returned by e-mail or post (participant choice). GP visits and infections were abstracted from GP records. Hospital attendances, cancer diagnoses and deaths were ascertained by linkage to Hospital Episode Statistics and national registration through NHS Digital. Setting: GP practices in England. Participants: Recruitment opened in October 2013 and closed in January 2015. A total of 1615 registered patients aged 65–84 years were randomised: 407 to vitamin D and 421 to no treatment in open practices; 395 to vitamin D and 392 to placebo in blind practices. Interventions: There was a 24-month treatment period: 12 monthly doses (100,000 IU of vitamin D3 or placebo as 5 ml oily solution) were posted after randomisation and at 1 year (100,000 IU per month corresponds to 3300 IU per day). Reminders were sent monthly by e-mail, text message or post. Main outcome measures: Recruitment, compliance, contamination and change in circulating 25-hydroxyvitamin D [25(OH)D] from baseline to 2 years. Results: Participation rates (randomised/invited) were 15.0% in open practices and 13.4% in double-blind practices (p = 0.7). The proportion still taking study medication at 2 years was 91.2% in open practices and 89.2% in double-blind practices (p = 0.4). The proportion of control participants taking > 400 IU vitamin D per day at 2 years was 5.0% in open practices and 4.8% in double-blind practices. Mean serum 25(OH)D concentration was 51.5 nmol/l [95% confidence interval (CI) 50.2 to 52.8 nmol/l] with 82.6% of participants < 75 nmol/l at baseline. At 2 years, this increased to 109.6 nmol/l (95% CI 107.1 to 112.1 nmol/l) with 12.0% < 75 nmol/l in those allocated to vitamin D and was unaltered at 51.8 nmol/l (95% CI 49.8 to 53.8 nmol/l) in those allocated to no vitamin D (no treatment or placebo). Conclusions: A trial could recruit 20,000 participants aged 65–84 years through 200 GP practices over 2 years. Approximately 80% would be expected to adhere to allocated treatment (vitamin D or placebo) for 5 years. The trial could be conducted entirely by e-mail in participants aged < 80 years, but some participants aged 80–84 years would require postal follow-up. Recruitment and treatment compliance would be similar and contamination (self-administration of vitamin D) would be minimal, whether control participants are randomised openly to no treatment with no contact during the trial or randomised double-blind to placebo with monthly reminders. Trial registration: Current Controlled Trials ISRCTN46328341 and EudraCT database 2011-003699-34. Funding: This project was funded by the National Institute for Health Research Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 24, No. 10. See the NIHR Journals Library website for further project information.