Metacognition has been accorded a role in both mathematical problem solving and in the learning of mathematics. There has been consistent advocacy of the need for the promotion of metacognitive activity in both domains. Such advocacy can only be effective if the advocated process is well understood. In this paper we have four goals: to describe a "multi-method" technique developed to study student mathematical metacognition; to set out the structural elements and configuration of a coherent model of metacognition in the domain of mathematical problem solving; to report on the empirical utility (and validity) of this model; and, to report the insights into student mathematical metacognition arising from the research. Multi-Method Interview Tasks are appended. (Contains 4 figures.)