This paper describes the development and initial validation of the Miller Screening for Preschoolers (MSP). A pilot edition of the test was administered to 174 preschool children in the state of Colorado. Of these, 39 had been previously identified as developmentally delayed. On the basis of preliminary analyses, test items were discarded if they were insensitive to differences between the at-risk group and the not-at-risk group or if they were nonsignificantly related to developmental age trends; if they were only weakly related to the domain total score; if they were redundant with other items within a domain; if they were too easy or too difficult for any given age; if they were rated low by the testers on administration and scoring issues; or if the test materials were deemed to be too expensive. This paper presents final statistics for the selected screening test items and discusses implications for the design, construction, and validation of developmental tests.