This study examined the need to develop reliable and valid procedures for screening large populations for possible speech and language problems and the use of the Fluharty Preschool Speech and Language Screening Test (FPSLST) as such a device. The test was administered to 260 preschool children, ages three to six. There were 166 Blacks and 94 Whites in the sample with a mean age of 4.1 years. Subjects came from two rural counties in the southern United States. Results indicated that age was positively related at a moderate level to the total subtest scores and that the overall strength of the race variable as expressed by phi values was relatively low. Analyses of step-wise discrimination and item difficulty raised questions concerning the usefulness and appropriateness of many test items. The deletion of items on the language-related subtests did not appear to save enough testing time to warrant their removal and resulting mis-classification of subjects. Multiple-regression procedures performed on the three language subtests revealed that the power of any single subtest to predict other subtests, or combination of two subtests to predict scores on a third subtest, accounted for such small degrees of variance that the continued use of all three subtests was deemed necessary. As for the internal consistency reliability of the three language subtests, only the Repetition subtest showed acceptable item homogeneity. (PN)