We consider the welfare consequences of nudges and other behavioral economic devices to encourage exercise habit formation. We analyze a randomized trial of nudged exercise commitment contracts in the context of a time-inconsistent intertemporal utility maximization model of the demand for exercise. The trial follows more than 4,000 people seeking to make exercise commitments. Each person was randomly nudged towards making longer (20 weeks) or shorter (8 weeks) exercise commitment contracts. Our empirical analysis shows that people who are interested in exercise commitment contracts choose longer contracts when nudged to do so, and are then more likely to meet their pre-stated exercise goals. People are also more likely to enroll in a subsequent commitment contract after the original expires if they receive a nudge for a longer duration initial contract. Our theoretical analysis of the welfare implications of these effects shows conditions under which nudges can reduce utility even when they succeed in the goal of promoting habitual exercise.