The 1992-93 school year saw the second statewide implementation of the Vermont portfolio-assessment program, and RAND continued its ongoing evaluation of the program's implementation, effects, and data quality. While the first year's study found evidence of the impact of the assessment program and low reliability of portfolio scoring, this year's evaluation focused only on the quality of the assessment data. The program was altered in many ways in 1992-93, although it continued its emphasis on portfolios, and this further evolution resulted in a clear increase in the reliability with which mathematics portfolios were scored. While more detailed scores are not as reliable, simple total scores for each portfolio showed moderately high reliability of scoring, particularly for the eighth grade. The reliability of writing portfolio scores did not improve substantially and was considerably lower than that for mathematics. The evaluation finds that reporting scores for supervisory unions (administrative units) is not especially informative and has large margins of error. The improvement in the reliability of scoring means that it is important to begin to focus on validity. (SLD)