The aim of our research was to study the effects of metacognition training on the performance of older adults solving mathematical word problems. A further aim was to study the links between metacognition and executive function. Thirty-two subjects aged over 60, divided into an experimental and a control group, took part in this study which involved five training sessions. Results show that metacognition training enhanced the two metacognition components (knowledge and skills) and the mathematical problem-solving capacities of the participants. They also suggest that the use of metacognition by older people to solve mathematical word problems is supported by executive functions (updating, shifting) and particularly by processing speed.