The objectives of this study are (1) to design an accurate method for linking newborn screening (NBS) and state birth certificate databases to create a de-identified study database; (2) To assess maternal cytomegalovirus (CMV) seroprevalence by measuring CMV IgG in newborn dried blood spots; (3) To assess congenital CMV infection among newborns and possible association with preterm birth. NBS and birth databases were linked and patient records were de-identified. A stratified random sample of records based on gestational age was selected and used to retrieve blood spots from the state NBS laboratory. Serum containing maternal antibodies was eluted from blood spots and tested for the presence of CMV IgG. DNA was extracted from blood spots and tested for the presence of CMV DNA. Analyses were performed with bivariable and multivariable logistic regression models. Linkage rates and specimen collection exceeded 98% of the total possible yielding a final database with 3,101 newborn blood spots. CMV seroprevalence was 91% among Black mothers, 83% among Hispanic mothers, 59% among White mothers, and decreased with increasing amounts of education. The prevalence of CMV infection in newborns was 0.45% and did not vary significantly by gestational age. Successful methods for database linkage, newborn blood spots collection, and de-identification of records can serve as a model for future congenital exposure surveillance projects. Maternal CMV seroprevalence was strongly associated with race/ethnicity and educational level. Congenital CMV infection rates were lower than those reported by other studies and lacked statistical power to examine associations with preterm birth.