The cloze technique is explored as a basis for measuring the skill of non-native speakers in handling English prepositions. Three groups of subjects are tested on an English cloze passage in which only prepositions have been deleted: the three groups consist of 19 native speakers of English (G-I), 53 native speakers of Turkish (G-II), and 110 foreign students entering UCLA from a wide variety of language backgrounds (G-III). An item analysis and a response frequency analysis are computed along with various test statistics for each group. The data suggest that a cloze test of prepositions works best with students from a variety of backgrounds (G-III) but is also useful for a group from a homogeneous language background (G-II). The response frequency analysis reveals that certain intralingual confusion occurs regardless of language background; other errors for G-II seem best explained by a contrastive analysis based on translation equivalents.