Zero-tillage (ZT) is a proven technology for sustainably enhancing wheat productivity in the densely populated Indo-Gangetic Plains (IGP). However, adoption remains modest, especially in the less productive East where farms are small, poverty is widespread, and ZT is relatively new and accessed mostly via custom-hiring services. To inform policy-makers and derive extension recommendations, we use a unique panel dataset from 961 wheat-growing households in Bihar to explore the dynamics of ZT adoption and its determinants over a three-year period, accounting for social network effects and access to ZT services. Using a heckprobit approach we estimate determinants of ZT awareness and -use in 2012 and 2015, correcting for non-exposure bias. We apply a multinomial logit model to identify determinants of early adoption, recent adoption, non-adoption, and dis-adoption. We find that a strong initial scale bias in knowledge and use of ZT declined substantially over the subsequent three-year period. Land fragmentation replaced total landholding size as a significant adoption determinant, education and caste mattered less, and the effect of farmers social networks increased. We conclude that the deployment of ZT through private-sector service providers can lead to relatively socially inclusive outcomes and recommend that service provision be fostered through appropriate policies. Acknowledgement : We gratefully acknowledge the willingness of the interviewed farm households to participate in the surveys. We thank USAID and the Bill & Melinda Gates Foundation for funding this research through the Cereal Systems Initiative for South Asia (CSISA) under the CGIAR Research Program on Wheat.