One of the most ambitious attempts to assess students' understanding of mathematics was carried out in England between 1974 and 1979. The results of this study indicate that there is a need in an overall assessment program to evaluate students' algebraic understanding as well as to measure their achievement in algebraic manipulations and problem solving. The present study is a modified replication of the English study. In 1977, 1981, and 1985 the British Columbia Ministry of Education carried out large scale assessments of mathematics achievement. Questions were designed to assess students' levels of algebraic understanding ranging from basic simplification to difficult relationships. The responses were classified by coders into one of nine categories, for example, correct, correct but weak, ambiguous, letter ignored. This scheme was extended further by coding the specific responses. It was thus possible to obtain frequency counts of each individual response as well as to assess overall error rates. A reliability check on coding yielded a 95.8 percent agreement rate on the 425 returned booklets. Four criteria were used to determine sample representativeness: gender, type of mathematics course, future mathematics plans, and performance on three multiple-choice items. Results indicate that performance of students on the three multiple-choice items in the sample was significantly higher than the population. Possible causes for this result are discussed. The appendix contains the British Columbia Mathematics Assessment (1985) for grade 10. (PN)