Along with the rise of digital games over the past decades came an increased interest for using games for other purposes than entertainment. Although a few successes are known, much research seems to suggest little evidence for games’ advantages. Existing literature claims that more studies are needed that investigate the effective design and use of games and especially studies that are comprehensive, rigorous, and innovative. To contribute to this emerging field, the author investigated the case of Levee Patroller. The target audience of the game, levee patrollers, are considered the “eyes and ears” of the Dutch water authorities. They inspect levees and report any risks they encounter. Similarly, in the game players have to find all virtual failures in a region and report these. If they do not find the failures in time or report them incorrectly, it could result in a levee breach that floods the whole virtual region. Using this game an innovative game-based training was set up to prove its effectiveness in training inspection knowledge and skills, and to understand the contributing factors. In total 147 levee patrollers from 3 water authorities in the Netherlands participated in a structured 3-week training which was evaluated using a quasi-experimental design with a mix of quantitative and qualitative methods. The results highlight a successful training. Clear evidence was found that the patrollers improved on their inspection knowledge and skills. But because how players performed in the game is most crucial for the game’s success as a training tool, future research should consider game design, data, and performance more elaborately.