Summary: The purpose of the study was to investigate the efficacy of using faded worked examples presented in matrices with problem structure variability to enhance learnersâ€™ ability to recognize the underlying structure of the problems. Specifically, this study compared the effects of matrix-format versus linear-format faded worked examples combined with equivalent problem structure versus contrast problem structure on learning. A total of 113 undergraduate students recruited from campus were randomly assigned to one of the four experimental conditions formed by a $2\times 2$ factorial design. The results revealed three significant interactions on accuracy of anticipations, near transfer and medium transfer, suggesting that matrices foster learning when they contain contrast-structure problems but not with equivalent-structure problems.