The purpose of this study was to understand the factors that affected teachers' adoption of game-based learning (GBL). In particular, this study paid special attention to the non-adopters who did not have any intentions to adopt GBL in the future and examined how these non-adopters differed from the early adopters, adopters and likely adopters. Based on Rogers' diffusion of innovation model, this study identified three factors - lifestyles, perceived attributes of GBL, and demographics - as significant predictors for adoption. Two methods were adopted by this study: First, this study conducted 15 intensive interviews to construct questions regarding the perceived attributes of GBL. Second, using a purposive sampling method, this study mailed 347 questionnaires to teachers in Taiwan's elementary schools and obtained 307 valid questionnaires. Two conclusions were drawn from the data analysis: (1) lifestyles and perceived attributes exerted a strong effect, whereas demographics only had a moderate effect in discriminating non-adopters from adopters and likely adopters; (2) the non-adopters in this study were more likely to be males instead of females, to have more teaching experience, to be older, to have no experience with video games, and to be better educated.