Pupils’ poor achievement in mathematics has recently been aconcern in many Western countries. In order to address this is-sue, it has been proposed to teach chess in schools. However,in spite of optimistic claims, no convincing evidence of the ac-ademic benefits of chess instruction has ever been provided,because no study has ever controlled for possible placebo ef-fects. In this experimental study, a three-group design (i.e., ex-perimental, placebo, and control groups) was implemented tocontrol for possible placebo effects. Measures of mathematicalability and metacognitive skills were taken before and after thetreatment. We were interested in metacognitive skills becausethey have been claimed to be boosted by chess instruction, inturn positively influencing the enhancement of mathematicalability. The results show that the experimental group (partici-pants attending a chess course) achieved better scores in math-ematics than the placebo group (participants attending a Gocourse) but not than the control group (participants attendingregular school lessons). With regard to metacognition, no dif-ferences were found between the three groups. These resultssuggest that some chess-related skills generalize to the mathe-matical domain, because the chess lessons compensated for thehours of school lessons lost, whereas the Go lessons did not.However, this transfer does not seem to be mediated by meta-cognitive skills, and thus appears to be too limited to offer ed-ucational advantages.