Students in grades 1, 4, 7, and 10 were tested in a two-part investigation of simple and complex mental addition (with college students as a reference point). One session involved a normal reaction time task in which children made true/false judgments about a series of addition examples. The other session involved a verbal protocol interview, the results of which are compared to the time-based results for an examination of children's mental addition performance. The hypothesized shift from counting to memory retrieval seemed to occur at about fourth grade, although older students continued to rely on immature processing for certain example types. A subgrouping of the fourth graders by mathematical ability was possible, yielding a strong effect on the reaction time measure. (Author/MNS)