This study used ecological momentary assessment (EMA) to test the association between activity, location, and social company contexts with cancer caregivers' in-the-moment affect to identify precisely when and where to deliver psychological interventions for caregivers. Current cancer caregivers (N = 25) received 8 EMA prompts per day for 7 consecutive days. At each prompt, caregivers reported their current positive affect and negative affect, as well as what they were doing, where they were located, and who they were with. Multilevel logistic regressions tested the associations between caregivers' contexts with their own person-mean-centered state (concurrent momentary level) and trait (overall weekly average) positive or negative affect. Caregivers reported lower state negative affect, as well as higher state positive affect, when socializing (ps < .001), when at a public location (ps < .03), and when around their friends, family, spouse/partner, or care recipient (i.e., person with cancer, ps < .02), relative to when not endorsing the context. Caregivers also reported lower state negative affect when eating/drinking or engaging in leisure (ps < .01; but no parallel effects for state positive affect). Caregivers reported higher state negative affect while working, when at their workplace, or when around work colleagues (ps < .001) and lower state positive affect when at home or alone (ps < .03). Results suggest the pertinence of a behavioral activation framework to mitigate the emotional strain of caregiving. Interventions that facilitate caregivers' ability to socialize with a range of friends and family, including their loved one with cancer, outside of the home may have the strongest positive emotional impact.