Three methods of equating performance assessments are compared. Equipercentile equating, item response theory (IRT) partial credit model, and IRT graded response model methods were applied to the same data using an anchor test design. Comparisons are based primarily on the resulting raw score-to-raw score tables and the effects these tables have on the classification of students in terms of two cut scores. The design entailed administering the performance assessment and a norm-referenced test, the anchor test, to each student. Matched data from between 4,950 and 7,820 students were analyzed at grades 3, 5, 8, and 10 in 1993 and 1994. The norm-referenced tests were the Iowa Tests of Basic Skills at the lower grades, and the Tests of Achievement and Proficiency for grade 10. All three methods produced relatively consistent results. In examining the percentage of students classified into performance levels in 1993 versus 1994 with the three equating methods, the equipercentile method produced the most consistent results. Both the partial credit and graded response methods had more deviation in the classification rates with the graded response model being more inconsistent. (Contains two figures, three tables, and seven references.) (SLD)