We combine exogenous variation in temperature at the county-day level in the U.S. with daily time use data to examine the effect of temperature on joint time use. We show that low temperatures reduce time spent with friends but increase time spent with family. Conversely, high temperatures increase time alone but reduce time with family. We also provide evidence of the effect of temperature on joint time use being location-dependent. We rationalize this finding using a model in which the chosen time allocation is the outcome of a dual-self decision process with an indoor and an outdoor self. The two selves have different tastes for time alone, time with family, and time with friends. Weather conditions can change the influence of each self, and thereby the corresponding preferences for joint time use. We test the predictions of the model empirically by drawing on methods from the household economics literature. The test results support the hypothesis that weather affects joint time use insofar it affects where the activities take place.