A study investigated the retention and loss of school-learned foreign language skills. Subjects were 150 students of Dutch secondary schools with 4 or 6 years of French language instruction and 0, 2, and 4 years of disuse. Nine tests of general language proficiency, listening and reading comprehension, phonology, vocabulary, and grammar were administered. Analysis of the results indicates that: (1) some integrative tests may be more sensitive than others to changes in universal language proficiency; (2) the multiple choice cloze test used here was a more valid indicator of French proficiency than the listening and reading comprehension tests; (3) the cloze test failed to reveal loss of vocabulary or grammar; and (4) contrasts in the design-dependent and design-independent analyses imply a need to separate validity and reliability studies of language tests from their application in research or education. A brief list of references is appended. (MSE)