Tested the effects of 2 levels of commitment to grass cycling (GC; i.e., not bagging grass clippings). During 4 wks of baseline, data were collected for all houses in 3 homogenous neighborhoods. 558 households were found to bag grass clippings. Neighborhoods were then assigned to commitment only (CO), commitment/agent (CA), or control (C) conditions. In the CO condition, 40 Ss made a commitment to GC. In the CA condition, 40 Ss made a commitment to GC and to promote that behavior to their neighbors. 40 Ss not contacted served as Cs. Ss from the CA condition had grass bags present significantly less often than either the CO or C Ss. There was a diffusion effect in which the neighbors of CA Ss showed significantly more GC than C Ss and this effect continued to increase through the 1-yr follow-up measure. CO behavior did not differ significantly from that shown by C Ss.