This report describes the development, operation, maintenance, and future prospects of the item banks pioneered by the Portland (Oregon) School District. At the time of this report, there were 3,500 mathematics, 2,200 reading, and 2,300 language usage items calibrated under the fixed parameter model of item response theory (IRT) for Grades 3-8. The advantages of common scaling of items through IRT methods are discussed, along with particulars for writing new items and conducting controlled field tests. Ongoing concerns include: (1) choosing the best IRT model (one-parameter, used currently, versus two- or three-parameter models); (2) adapting to curriculum dynamics, while preserving item comparability; (3) promoting curriculum alignment of test instruments; (4) making this complex system credible to those outside the testing research field. Twelve quality control guidelines are offered, covering proper printing of tests, test administration procedures, accurate coding and scoring, and retesting criteria. New directions include a pilot project in computerized adaptive testing at elementary and secondary levels and the development of large pools in science, guided by analysis of underlying constructs. The report provides names and addresses of those to contact for access to the item bank system. (LPG)