This paper aims to improve the design methods for serious games (games for learning) by identifying a set of well-established pedagogical misconceptions and presenting design guidelines to avoid these. It analyses the pedagogical principles and models that are commonly used in serious game design, and contrasts these with evidence and advances in instructional psychology and instructional design research. The paper particularly focuses on (1) the concept of experience-based learning, which many serious games comply with, (2) the concept of learner motivation, which most games strongly claim to support, and (3) the score systems that many games use to track and display progress. Structural design weaknesses are exposed and countered with a large body of research evidence from the literature. A set of practicable design guidelines are presented that help to avoid the pedagogical flaws and contribute to improving the design methods for serious games.