A problem-solving program supplemented by calculators in one treatment group was conducted in 63 grade 7 classes with about 1350 students. Teachers were provided with problems correlated with textbooks, and instruction for teaching problem-solving strategies. School districts provided calculators and problem-solving materials. Pretest scores indicated equality of treatment groups. On mid-year tests, the problem-solving group (P) performed significantly higher than the control group (C) on both traditional and "strategy" problems. On year-end tests, the problem-solving group using calculators (CP) performed better than the C group. The CP group scored higher on attitude toward problem solving than the other groups in the district with greatest participation. On every problem-solving test, at least one of the problem-solving groups, CP or P, scored higher than the C group. On a test of more difficult problems, the trend favored the P group, but not significantly. Sex differences were negligible. There were no differences on a paper-and-pencil computation test, indicating that students using calculators all year retained computational skills. The study shows the feasibility of teaching problem-solving strategies and using calculators in grade 7. Similar studies and implementation are suggested for earlier grades. (Author)