The dimensional structure of the 1987 Advanced Placement (AP) French language examination was tested in four populations using a series of confirmatory linear factor analysis models. To mitigate problems with the linear factor analysis of multiple choice items, the linear factor analysis of item parcel scores, made of small mutually exclusive collections of items hypothesized to measure the same underlying dimension, was used. Six confirmatory factor analysis models were tested with each of five data samples. Two contained high school AP candidates with no out-of-school French language experience (n=1,500 each), and a third consisted of candidates who had spent a significant amount of time in a French-speaking country (n=1,418). A fourth sample contained 477 AP candidates who regularly spoke or heard French at home, and the final sample contained 302 students with no out-of-class experience enrolled in third year college French. In all samples the examination appeared to measure four major dimensions associated with listening, reading, writing, and speaking. For those without out-of-school experience, the examination displayed invariance of factor loadings and errors of measurement. Factor structures were similar for groups with similar out-of-school French language experience. (Contains 15 tables, 1 figure, 2 tabular appendixes, and 13 references.) (SLD)