The purpose of this study was to determine differences in achievement of fifth-grade students in 16 classes taught two of four mathematical algorithms by one of four instructional strategies. Computational skills and ability to extend the algorithm were tested. The instructional strategies were: pattern, algebraic, pattern followed by algebraic-type justification, and algebraic followed by pattern-type justification. Two algorithms were classified as "simple": multiplication of fractions and mixed numbers and comparison of fractions by cross-products. The other two were conversion of fractions to decimals and calculation of the square root of fractions: these were listed as complex. Treatment groups were taught one simple and one complex algorithm. No significant differences were found for the cases of all four algorithms tested between the comparisons of the strictly pattern groups and the strictly algebraic groups. There was evidence to indicate algebraic groups tended to do better on simple algorithm extension tests than did the pattern groups; this was reversed for computation tests. No trend was evident in terms of performance of groups on complex algorithm computation tests. Results also indicated that the algebraic-followed-by-pattern students performed better, in general, on the extension tests. (Author/JP)