States and districts need ways of measuring principal performance that correctly identify effective principals. Unfortunately, existing research offers little guidance to policymakers on which types of performance measures provide valid information about principals' contributions to student achievement. States have therefore had to develop principal performance measures without clear evidence that these measures accurately identify effective principals. This study examined four alternative measures of principal performance. "Average achievement" used only information about students' end-of-year achievement without taking into account the students' past achievement. In contrast, "school value-added" accounted for students' own past achievement by measuring their growth--specifically, the extent to which student achievement growth at a school differed from average growth statewide for students with similar prior achievement and background characteristics. Two other measures in this study took into account the schools' prior performance to avoid rewarding or penalizing principals simply for being assigned to schools that had better or worse characteristics. "Adjusted average achievement" and "adjusted school value-added" credited principals if their schools' average achievement and value-added, respectively, exceeded predictions for the average principal, given the schools' past performance on those same measures. To assess each measure's predictive validity, the study conducted two sets of analyses using student and principal data from 2007/08-2013/14 in the entire state of Pennsylvania. First, the study assessed the extent to which ratings from each measure are stable--that is, remain consistent over time--by examining the association between principals' ratings from earlier and later years. Stability was important to measure because only stable parts of a rating have the potential to contain information about principals' future performance; unstable parts reflect only transient aspects of their performance. Second, the study examined the relationship between the stable part of a principal's rating and his or her contributions to student achievement in future years. To do so, the study carried out a benchmark approach to obtain the most rigorous available measure of principals' contributions--but one that was available for only a subset of principals. For the benchmark approach, the study team calculated the change in student achievement at a school when one principal replaced another to determine how the successor's contribution differed from that of the predecessor. The study then compared the stable parts of the ratings from each of the four measures to results from the benchmark approach in a future year. Using the results of both analyses, the study summarized each measure's predictive validity by simulating its accuracy for predicting principals' contributions to student achievement in the following year. A measure could have high predictive validity only if it was highly stable between consecutive years (from the first analysis) and its stable part was strongly related to principals' contributions to student achievement (from the second analysis). The study assessed the predictive validity of single-year ratings and ratings averaged across three years. The study had the following key findings: (1) The two performance measures in this study that did not account for students' past achievement--average achievement and adjusted average achievement--provided no information for predicting principals' contributions to student achievement in the following year; (2) The two performance measures in this study that accounted for students' past achievement--school value-added and adjusted school value-added--provided, at most, a small amount of information for predicting principals' contributions to student achievement in the following year; and (3) Averaging performance measures across multiple recent years did not improve their accuracy for predicting principals' contributions to student achievement in the following year. The following are appended: (1) Data used in the study; (2) Technical details of the principal performance measures that the study assessed; (3) Technical details of methods for examining the stability of principal performance ratings; (4) Technical methods for assessing the predictive validity of principal performance measures; (5) Study sample for assessing predictive validity; and (6) Supplemental findings.