We use data from the evaluation sample of Mexico's Food Assistance Program (PAL) to study whether including the attendance at health and nutrition classes among the requirements for receiving a transfer affects the health behavior of adults living in localities targeted by the program. The experimental trial has four different treatment types, randomly assigned to four groups of localities, one of which receives the in-kind transfer without the requirement to attend any health or nutrition sessions. Adult women living in localities where the in-kind transfer is conditional on class attendance display a significantly better health behavior than those living in localities where it is not. There is no significant evidence of changes in health outcomes among men. (Contains 3 figures, 15 tables, and 29 footnotes.)