Early childhood special educators recognize the necessity of establishing indices of reliability and validity for instruments that provide an index of developmental status. Many such instruments present little empirical evidence regarding psychometric integrity, particularly for a non-normative sample. The 341-item Battelle Developmental Inventory (BDI) is a commonly used instrument in early childhood special education research, but previously established indices of reliability and validity for this instrument were derived from samples of non-disabled children. This study investigated the item reliability and construct validity of the BDI for 78 children with severe disabilities. Data were collected as part of a longitudinal nationwide study investigating the efficacy of early intervention. Subjects were tested annually over a 5-year period with the BDI as one of several dependent measures, resulting in 265 cases across 8 age groups ranging from 11 months to 75 months. A generalizability study was conducted to establish the reliability of items; factor analytic procedures were used to determine construct validity. Results indicate that indices of reliability and validity differ from those obtained for the normative sample. Implications for use of the BDI and other developmental status measures in early childhood special education are discussed. (16 references; 14 data tables) (Author/SLD)