OBJECTIVE: To assess factors related to recorded vaccine uptake, which may confound the evaluation of vaccine impact. METHODS: Analysis of documented vaccination histories of children under 5 years and demographic and socio-economic characteristics collected by a demographic surveillance system in Karonga District, Malawi. Associations between deviations from the standard vaccination schedule and characteristics that are likely to be associated with increased mortality were determined by multivariate logistic regression. RESULTS: Approximately 78% of children aged 6-23 months had a vaccination document, declining to <50% by 5 years of age. Living closer to an under-5 clinic, having a better educated father, and both parents being alive were associated with having a vaccination document. For a small percentage of children, vaccination records were incomplete and/or faulty. Vaccination uptake was high overall, but delayed among children living further from the nearest under-5 clinic or from poorer socio-economic backgrounds. Approximately 9% of children had received their last dose of DPT with or after measles vaccine. These children were from relatively less educated parents, and were more likely to have been born outside the health services. CONCLUSIONS: Though overall coverage in this community was high and variation in coverage according to child or parental characteristics small, there was strong evidence of more timely coverage among children from better socio-economic conditions and among those who lived closer to health facilities. These factors are likely to be strong confounders in the association of vaccinations with mortality, and may offer an alternative explanation for the non-specific mortality impact of vaccines described by other studies.