The purpose of this study was to take an initial step toward developing sound and versatile predictive instruments (i.e., benchmark measures) in reading, writing, and math that can be used to assess high-school students' academic performance in a manner that will not only predict scores on high-stakes tests but will also be amenable to repeat administrations. Data were collected on 200 ninth-graders (6% with disabilities) enrolled in 8 math courses with a difficulty level ranging from remedial math to algebra. The sample for the reading benchmark assessment included 235 ninth-graders (33% were English language learners and 2% had disabilities), enrolled in 14 general education language arts courses. The sample for the writing benchmark assessment included 162 students, a subset of the reading sample. Results for the reading comprehension benchmark test indicated good levels of internal consistency. Additionally, variability in performance related to the sixth-grade reading passage was found to have the strongest predictive strength to overall performance variability on both SAT9 and HSEE reading sub-scales. For the mathematics benchmark test results, internal consistency estimates fell within an acceptable range. For the writing benchmark test, adequate levels of interrater reliability were established on the individual scoring criteria. (Contains 23 references and 7 tables.) (Author/CR)