Holds that interrater agreement for journal manuscript reviews has seemed unacceptably low because it has been assessed using techniques such as the intraclass correlation, which compares error variance with the variance due to manuscripts. Describes and recommends an alternative approach for computing interrater agreement. (GC)