The main objective of this study is to provide insight on how Indonesian farmer preferences for crop attributes influence their adoption decisions. Results from a Latent Class (LC) cluster analysis, using the individual scores for each of the Best-Worst (BW) scaling attributes, indicate there are four clusters of farmers, each distinct in their relative preferences for crop attributes and socio-demographic characteristics. The multinomial endogenous treatment regressions show that preference cluster effect varies across models. For the binary adoption model, we find an insignificant preference cluster effect. We find a significant preference cluster effect both for the intensity of adoption and the timing of adoption models. The effects of farmers’ crop preference cluster, however, are different across those models. The findings allow more targeted programming to encourage farmers to adopt high-value crops that have a high probability of offering benefits for farmers.