We conducted a randomized controlled trial testing the effect of modest incentives to attend the gym among new members of a fitness facility, a population that is already engaged in trying to change a health behavior. Our experiment randomized 836 new members of a private gym into a control group, receiving a $30 payment unconditionally, or one of 3 incentive groups, receiving a payment if they attended the gym at least 9 times over their first 6 weeks as members. The incentives were a $30 payment, a $60 payment, and an item costing $30 that leveraged the endowment effect. These incentives had only moderate impacts on attendance during members' first 6 weeks and no effect on their subsequent visit trajectories. We document substantial overconfidence among new members about their likely visit rates and discuss how overconfidence may undermine the effectiveness of a modest incentive program.