Potential health risks for informal caregivers have been hypothesised to be partly related to adverse changes in health-related behaviour, but evidence is limited. We examined whether smoking, drinking, eating, physical activity or leisure pursuits change in relation to co-resident or out-of-home caregiving (for someone outside the household), and if associations differ by sex, educational attainment, and welfare state typology. We conducted a longitudinal study using six waves of the Survey of Health, Ageing and Retirement in Europe, collecting data repeatedly from 2004 to 2017 for adults aged 50 years and older living in 17 European countries (57,962 individuals). To control for measured and unmeasured within-individual time-invariant confounders, we used fixed effects logistic models to analyse the repeated measures of caregiving, behaviour and covariates and estimated odds ratios (OR) with 95% confidence intervals (95%CI). Among male participants, unhealthy eating increased while smoking decreased [ORs 1.26 (95%CI 1.01-1.58) and 0.53 (0.36-0.78), respectively] in survey waves in which they provided co-resident care, compared with the waves that they did not. Among female participants, there was little change in behaviour between waves with and without co-resident caregiving. When providing out-of-home care, lacks of physical activity and leisure pursuits declined. But in the same time, drinking increased both men and women, and especially among individuals with lower educational attainment and residing in non-Nordic countries. To conclude, overall, increased drinking when providing out-of-home care was most consistent, especially among individuals with lower educational attainment and residing in non-Nordic countries. Otherwise, the associations varied by the type of care, behaviour and population subgroups. These findings indicated that not all caregivers are susceptible to behavioural changes, and that not all behaviour may be involved similarly in linking caregiving to health risks. This opens possibilities to target specific behaviour and groups to prevent adverse changes in health behaviour in caregivers.