BACKGROUND Obesity contributes to disability in older adults, and this is offset by weight loss and exercise. Very Low Calorie Diets (VLCDs) achieve rapid weight loss; however, these have not been rigorously evaluated in older people. METHODS A randomized trial was conducted from August 2012 through December 2015. The intervention was 12 weeks of thrice weekly exercise combined with either healthy eating advice (Ex/HE), hypocaloric diet (Ex/Diet), or VLCD (Ex/VLCD). Outcomes were physical function, measured by 6-minute walk test (6MWT) and De Morton Mobility Index (DEMMI). Other measures were body composition measured by Dual Energy X-Ray Absorptiometry, and nutritional parameters (albumin, vitamins B12 and D, ferritin and folate). RESULTS 36, 40, and 41 participants were randomized to Ex/HE, Ex/Diet, and Ex/VLCD, respectively. At 12 weeks, weight was reduced by 3.7, 5.1, and 11.1% (p < .01), respectively. Ex/VLCD had significant reduction in fat (16.8%), lean mass (4.8%), and bone mineral density (1.2%), but increased relative lean mass (3.8%). DEMMI improved by 14.25, 14.25, and 13.75 points in Ex/HE, Ex/Diet, and Ex/VLCD, respectively; however, there was no between-group difference (p = .30). 6MWT improved by 53.1, 64.7, and 84.4 meters in Ex/HE, Ex/Diet, and Ex/VLCD (p = .18). Post hoc stratification for gender and adjustment for initial physical function and type 2 diabetes only revealed significant between-group differences for men in the 6MWT, with improvement by 57.8, 77.8, and 140.3 meters in Ex/HE, Ex/Diet, and Ex/VLCD, respectively (p = .01). Improvements in nutritional parameters were seen in Ex/VLCD, but not in Ex/HE and Ex/Diet. The VLCD was well tolerated. CONCLUSIONS VLCDs have potential in the treatment of obesity in older persons; of particular benefit is improvement in nutritional status. The gait speed improvement observed in men warrants further investigation.