BACKGROUND: The increasing prevalence of frailty with age is becoming a public health priority in countries with ageing populations. Pre-frailty presents a window of opportunity to prevent the development of frailty in community-dwelling older adults. This study aimed to examine the effectiveness of a complex intervention that combined a nutrition-based intervention and a physical activity intervention, along with the effectiveness of each intervention individually, to reduce physical frailty in pre-frail older adults over 2 years. METHODS: In this single-blind, 2 x 2 factorial, randomised, controlled trial, we recruited pre-frail community-dwelling older adults in Aotearoa New Zealand via mail through general medical practices. To be eligible, participants had to be pre-frail according to self-reported FRAIL scores of 1 or 2, aged 75 years or older (or 60 years or older for Māori and Pacific Peoples), not terminally ill or with advanced dementia as judged by a general practitioner, able to stand, medically safe to participate in low-intensity exercise, and able to use kitchen utensils safely. Participants were randomly allocated to receive an 8-week Senior Chef programme (SC group), a 10-week Steady As You Go programme (SAYGO group), a 10-week combined SC and SAYGO intervention (combined group), or a 10-week social programme (control group), using computer-generated block randomisation administered through an electronic data capture system by local study coordinators. Assessors were masked to group allocation for all assessments. SC is a group-based nutrition education and cooking class programme (3 h weekly), SAYGO is a group-based strength and balance exercise programme (1 h weekly), and the social control programme was a seated, group socialising activity (once a week). Masked assessors ascertained Fried frailty scores at baseline, end of intervention, and at 6, 12, and 24 months after the programme. The primary outcome was change in Fried frailty score at 2 years. Intention-to-treat analyses were completed for all randomised participants, and all participants who had a high (≥75%) adherence were analysed per protocol. This study is registered at ANZCTR, ACTRN12614000827639. FINDINGS: Between May 12, 2016 and April 9, 2018, we assessed 2678 older adults for eligibility, of whom 468 (17%) consented and completed baseline assessment, with a mean age of 80·3 years (SD 5·1) and a mean Fried score of 1·9 (1·2); 59% were women. We randomly allocated these participants into the four groups: 117 in the SC group, 118 in the SAYGO group, 118 in the combined group, and 115 in the control group; 318 participants attended the 24-month follow-up: 89 in the SC group, 78 in the SAYGO group, 73 in the combined group, and 78 in the control group. At the 24-month follow-up, there were no differences in mean Fried scores between the intervention groups and the control group. No adverse events were reported. INTERPRETATION: The study did not find that the combined SC and SAYGO programme was effective in reducing frailty in pre-frail older adults. Although some short-term benefits were observed in each individual programme, there was no clear evidence of long-term impact. Further research is needed to evaluate combinations of group-based programmes for community-dwelling older adults to optimise their physical function. FUNDING: Health Research Council New Zealand and Ageing Well Challenge (Ministry of Business Innovation and Employment).