Serious games, or else educational games, for programming are considered to have a positive impact on learning programming. Specifically, serious games are considered to motivate students and engage them in playing and learning programming. However, more research is required in order to study their effects in learning programming, as well as their added value in comparison with typical teaching approaches. In this study the effects of teaching programming to lower secondary school students with the serious game CodeCombat and the typical teaching approach are compared. Specifically, fifty-nine students formed an experimental group that was taught programming with CodeCombat and a control group that was taught programming through lecturing and problem solving in Python with a typical programming environment. The study aimed to investigate whether a game like CodeCombat that is based on a text-based real programming language: improves students' performance in basic programming concepts; brings better learning outcomes in comparison with typical teaching methods; engages students' interest. Data was collected through a pre and post test, as well as a survey prior the intervention and another one based on the Technology Acceptance Model at the end. It was concluded that the experimental group performed better than the control group, but this difference was not found to be statistically significant. CodeCombat was evaluated positively in terms of its perceived ease of use and usefulness, as well as the attitude towards its use. The results were neutral in terms of students' behavioral intention to use CodeCombat, but were positive in using serious games for programming in general.