Informal Reading Inventories (IRI) are often recommended as instructionally relevant measures of reading. However, they have also been criticized for inattention to technical quality. Examination of reliability evidence in nine recently revised IRIs revealed that fewer than half report reliability. Several appear to have sufficient reliability for lower stakes decisions such as selection of classroom reading materials, but not for higher stakes purposes such as identification of reading difficulties. This article provides recommendations for improving IRI reliability and addresses the need for expanded guidelines for evaluating reliability, particularly for measures of score agreement. (Contains 1 table.)