The aims of this study were fourfold: to document the prevalence of language delays in a sample of at-risk 3 year olds; to assess the effectiveness of a home visiting program in preventing early language delays; to determine how often parents, pediatric providers, and home visitors identified early language delays; and to assess the effectiveness of a home visiting program in improving early identification of language delays. The Preschool Language Scale, Third Edition (PLS-3) was administered to 513 at-risk 3 year olds participating in a randomized trial of home visiting services. Families randomized to home visiting were expected to receive weekly to quarterly visits throughout the 3 years of this study. The content of home visits included teaching parents about child development, role-modeling parenting skills, and linking families to a medical home. Identification of delays was measured using structured parent interviews and review of primary care and home visiting records. At age 3 years, 10% of children had severe language delays, defined as scoring >or=2 SD below the national mean on the PLS-3, whereas 49% scored >or=1 SD below the national mean. No differences in prevalence were seen between children who did and did not receive home visiting. Among children with severe delays, 42% were identified by parents, 33% by pediatric providers, and 24% by home visitors. Among children with any delays, 24% were identified by parents, 25% by pediatric providers, and 17% by home visitors. No differences in rates of identification were seen between children who did and did not receive home visiting. Thus, while language delays were highly prevalent among these at-risk children, rates of identification were low, even among children with severe delays. Home visiting was not effective in either preventing language delays or improving early identification. This suggests that pediatric providers and home visiting programs need to reexamine their approaches to recognizing and intervening with early language delays.