BACKGROUND: Insufficient physical activity is a determinant of obesity and cardiovascular disease. Active travel to work has declined in high-income countries in recent decades. We aimed to determine which socioeconomic and demographic characteristics predicted switching to or from active commuting, whether switching from passive to active commuting (or the reverse) independently predicts change in objectively measured body-mass index (BMI), and to ascertain whether any association is attenuated by socioeconomic, demographic, or behavioural factors. METHODS: This study used longitudinal data from UK Biobank. Baseline data collection occurred at 22 centres between March, 2006, and July, 2010, with a repeat assessment at one centre (Stockport) between August, 2012, and June, 2013, for a subset of these participants. Height and weight were objectively measured at both timepoints. We included individuals present at both timepoints with complete data in the analytic sample. Participants were aged 40-69 years and commuted from home to a workplace on a regular basis at both baseline and follow-up. Two exposures were investigated: transition from car commuting to active or public transport commuting and transition from active or public transport to car commuting. Change in BMI between baseline and repeat assessment was the outcome of interest, assessed with bivariate and multivariate logistic regression models. FINDINGS: 502 656 individuals provided baseline data, with 20 346 participating in the repeat assessment after a median of 4·4 years (IQR 3·7-4·9). 5861 individuals were present at both timepoints and had complete data for all analytic variables. Individuals who transitioned from car commuting at baseline to active or public transportation modes at follow-up had a decrease in BMI of -0·30 kg/m2 (95% CI -0·47 to -0·13; p=0·0005). Conversely, individuals who transitioned from active commuting at baseline to car commuting at follow-up had a BMI increase of 0·32 kg/m2 (0·13 to 0·50; p=0·008). These effects were not attenuated by adjustment for hypothesised confounders. Change in household income emerged as a determinant of commute mode transitions. INTERPRETATION: Incorporation of increased levels of physical activity as part of the commute to work could reduce obesity among middle-aged adults in the UK. FUNDING: UK Medical Research Council.