This study compared the acquisition of an initial place value skill when presented in a concrete, semiconcrete, abstract teaching sequence to acquisition of the same skill when presented at the abstract level only. The 24 subjects were elementary and middle school students (ages 8-13) with learning disabilities who were randomly assigned to experimental and control groups. A 2x3 mixed design with one between (treatment) and one within (performance over time) group factor was used. A significant main effect for the treatment was found on the acquisition measure, suggesting that the concrete to abstract teaching sequence is more effective than abstract teaching alone for this concept with this population. The use of manipulatives and pictorial representations positively affected skill acquisition, maintenance, and retention, thus supporting at least as much instructional emphasis on concept teaching as on computation skills. No significant difference was found on the generalization measure. Appended are a sample posttest and a sample teaching script for a concrete lesson. (Contains 23 references.) (DB)