This paper presents arguments in favor of using partial dictation as a test of foreign language proficiency. In this kind of test, subjects listen to recordings of material in the foreign language and are required to fill in missing words in a written version of the recordings. Partial dictation is preferable to ordinary dictation in that: (1) it makes possible the use of different voices and speech situations; (2) it is more economical; and (3) the testing situation is more natural, being subject to fewer interruptions. In addition, partial dictation appears to be reliable and to correspond very well to other measures of foreign language proficiency; it is easy to construct, administer, and score; and it is useful, not only as a measure of listening comprehension, but as a global estimate of language proficiency. An incomplete analysis is presented of errors found in connection with this test, on the levels of phonology, lexicology, and grammar. Appendices contain two sample passages used in a partial dictation test, and a listing of average scores for each item, from a random sampling of 20 subjects. (Author/AM)