BACKGROUND: This study evaluated the utility of immunization registries in identifying vaccine refusals among children. Among refusers, we studied their socioeconomic characteristics and health care utilization patterns. METHODS: Medical records were reviewed to validate refusal status in the immunization registries of two health plans. Racial, education, and income characteristics of children claiming refusal were collected based on the census tract of each child. Health care utilization was identified using both electronic medical record and insurance claims. Within the immunization registries of two HMOs in the study, some providers use refusal and medical contraindication interchangeably, and some providers tend to always use "ever refusal." Therefore, we combined medical contraindication and refusal together and treated them all as "refusal" in this study. RESULTS: The immunization registry, compared to chart review, had negative predictive values of 85-92% and 90-97% for 2- and 6-year olds, and positive predictive values of only 52-74% and 59-62% to identify vaccine refusals. Refusers were more likely to reside in well-educated, higher income areas than non-refusers. Refusers had not opted out of health care system and continued, although less frequently for the age 2 and under group, to use services. CONCLUSION: Without enhancements to immunization registries, identifying children with immunization refusal would be time consuming. Since communities where refusers live are well educated, interventions should target these communities to communicate vaccine adverse events and consequences of vaccine preventable diseases.