Previous work has shown that the cognitive processes involved in mental arithmetic can bedecomposed into three stages: encoding, calculation, and production. Models of mental arithmetic hypothesize varying degrees of independence between these processes of encoding and calculation. In the present study, we tested whether encoding and calculation are independent by having participants complete an addition verification task. We manipulated problem size (small, large) as well as problem format, having participants verify equations presented either as Arabic digits (e.g., “3 + 7 = 10”) or using words (e.g., “three + seven = ten”). In addition, we collected trial-by-trial strategy reports. Though we found main effects of both problem size and format on response times, we found no interaction between the two factors, supporting the hypothesis that encoding and calculation function independently. However, strategy reports indicated that manipulating format caused a shift from retrieval based strategies to procedural strategies, particularly on large problems. We discuss these results in light of two competing models of mental arithmetic.