Drawing on the literature that posits heterogeneous influences of social networks on health behaviors, we tested whether different forms of participation in voluntary organizations predicted more or less alcohol and tobacco consumption over time. (Access preregistration at https://osf.io/guzem/) We used panel data from younger (aged 14–29 at baseline), middle-aged (aged 40–50), and older (aged 65–75) UK adults, Ns = 1280–9073, followed from 1991 to 2014. Annual measures of smoking included status and intensity. Frequency of pub attendance was assessed biennially between 1996 and 2008. In 2010 and 2013, more precise measures of alcohol consumption were available. We conducted two-level regression analyses for the outcomes measured more than twice and residual change analyses for other outcomes. Over time (within persons), there were no significant effects on smoking. Activity in voluntary organizations predicted slightly less frequent pub attendance in younger adults. In residual change analyses, activity in voluntary organizations decreased last-week ethanol consumption and risk of heavy episodic drinking in younger women. These effects pertained mainly to service-orientated organizations. In middle-aged adults, membership and attendance at meetings of voluntary organizations predicted slightly more frequent pub attendance. Residual change analyses showed volunteering to reduce the risk of heavy episodic drinking in middle-aged men. In older adults, few significant effects emerged. Between persons, all indicators of participation were associated with less smoking, whereas membership was associated with more and activity with less frequent pub attendance. Thus, most associations between participation in voluntary organizations and substance use reflected interindividual differences.