This study examines the assumptions underlying the scoring system of the Arizona Articulation Proficiency Scale: Revised. Twenty-one children between the ages of four years two months and six years 11 months were administered the Arizona Articulation Proficiency Scale: Revised and the Screening Deep Test of Articulation. The subjects' performance on the two tests was compared for the phones: [s], [l], [r], [t∫], [θ], [∫], [k], [f], and [t]. Results suggested that 1) the production of most sounds in only two contexts does not necessarily reflect the accuracy of production of those sounds in other contexts, and 2) for the sounds tested, the weightings assigned on the basis of their frequency of occurrence rather than the frequency with which they were misarticulated overestimated the extent of many articulation errors.