The convergent and discriminant validity of the Early Development Instrument (EDI), a teacher-rated assessment of children's "school readiness", was investigated in a multicultural sample of 267 kindergarteners (53% male). Teachers evaluations on the EDI, both overall and in five domains (physical health/well-being, social competence, emotional maturity, language/cognition, communication/general knowledge), were related to direct, child-based assessments of performance on two standardized measures of school readiness, and measures of phonological awareness and early social competence. Regression analysis indicated that together the four comparison measures accounted for 36% of variance in overall EDI scores, each making a significant and unique contribution. Results supported the convergent validity of overall EDI scores but not the discriminant validity of EDI domain scores. Moreover, correlations between EDI scores and comparison measures varied widely across teachers, suggesting considerable individual differences in teacher's ability to evaluate school readiness relative to direct, child-based assessments, and confirming that the EDI is more appropriate for deriving inferences at higher aggregated levels such as community or region. The validation of EDI domain scores remains an important challenge in future research.