Mourato, Fausto JoséSantos, Manuel Próspero dosMealha, Óscar and Madeira, Joaquim and Tércio, Daniel and Sousa Santos, Beatriz2021-06-182021-06-182021978-3-03868-153-3https://doi.org/10.2312/pt.20101177https://diglib.eg.org:443/handle/10.2312/pt20101177Automatic generation of game levels improves replayability and also allows content adaptation. One important aspect to take into account in the creation of any videogame is difficulty, in particular when it is possible to adapt content. However, defining difficulty is not a straight subject. In this paper we propose a metric for game difficulty in platform levels, mostly based on users' losing probability for each obstacle. This metric can be further used in automated processes that generate levels for this type of games, helping the process to recognize whether a level is suitable or not for a certain player. We also present some examples of the usage of this metric in commercial games.Human FactorsPlatform videogamesdifficulty measurement10.2312/pt.20101177173-180