We conducted a 14-week experimental study of 2 versions of a relatively comprehensive RC intervention that involved 50 classroom teachers, 15 tutors, and 116 children drawn in equal proportions from grades 3 and 5 in 13 schools in a large urban school district. Students were randomly assigned in equal numbers to the two tutoring conditions and a control group. Results indicated that students in the two tutored groups tended to perform comparably on all tests and to outperform controls (more so in grade 5 than grade 3) on near-transfer but not far-transfer measures of RC. This differential pattern of program effects for near- versus far-transfer measures raises questions about how tests of near-transfer and far-transfer are conventionally understood.