The objectives of the current study were to examine: (a) the effects of the Peer Mediation with Young Children (PMYC) program on mediators' and learners' mediation style in the context of maths teaching; and (b) the interaction effects of the intervention program, children's verbal ability, and maths level on pre-to post-intervention improvement in maths and mediation style. The sample was composed of 108 pupils (54 mediator--learner dyads), equally divided between an experimental and a control group. In each dyad, the mediator was from Grade 3 and the learner from Grade 2. Mediators and learners in each pair were matched in a counterbalanced design (2x2) according to their level of maths skills (high versus low). The experimental children received the PMYC program whereas the control children received general preparation for peer-assisted learning only. Following the PMYC intervention, all children participated in a peer-mediation condition where the mediators taught their younger peers how to solve maths problems. The interaction was videotaped for 25 minutes and analysed by the "Observation of Mediation Instrument." The main findings showed that the experimental children obtained significantly higher scores on all mediation criteria and improved their pre-to post-intervention scores in maths to a greater degree than did the control children. These findings are explained in relation to Vygotsky's and Feuerstein's theories in addition to previous findings on peer mediation. (Contains 4 tables and 1 figure.)