The instructional effectiveness of exploring computer-based simulation games is hypothesized to be low unless teaching functions are implemented. According to Klauer's (1985) framework for a theory of teaching, two varieties of instructional support were investigated in three experiments: (1) System-initiated adaptive advice and (2) learner-requested non-adaptive background information. Advice increased verbal domain knowledge, but decreased game performance (N = 64 Grade 7 students) - an effect replicated with university students (N = 38); advice had short-term effects, background information had long-term effects (N = 80 Grades 7-8 students). ATI-effects were observed. The results are discussed concerning instructional implications and cognitive theories of problem-solving, intelligence, memory overload and depth of information processing.