This study examined the influence of virtual manipulatives on different achievement groups during a teaching experiment in four fifth-grade classrooms. During a two-week unit focusing on two rational number concepts (fraction equivalence and fraction addition with unlike denominators) one low achieving, two average achieving, and one high achieving group participated in two instructional treatments (three groups used virtual manipulatives and one group used physical manipulatives). Data sources included pre- and post-tests of students' mathematical content knowledge and videotapes of classroom sessions. Results of paired samples t-tests examining the three groups using virtual manipulatives indicated a statistically significant overall gain following the treatment. Follow-up paired samples individual t-tests on the low, average, and high achieving groups indicated a statistically significant gain for students in the low achieving group, but only numerical gains for students in the average and high achieving groups. There were no significant differences between the average achieving student groups in the virtual manipulatives and physical manipulatives treatments. Qualitative data gathered during the study indicated that the different achievement groups experienced the virtual manipulatives in different ways, with the high achieving group recognizing patterns quickly and transitioning to the use of symbols, while the average and low achieving groups relied heavily on pictorial representations as they methodically worked step-by-step through processes and procedures with mathematical symbols. (Contains 2 tables.)