Estimates of immunization coverage in developing countries are typically made on a "card plus history" basis, combining information obtained from vaccination cards with information from mothers' reports, for children for whom such cards are not available. A recent survey in rural lower Egypt was able to test the accuracy of mothers' reports for a subset of children whose cards were not seen at round 1 of the survey but were seen a year later at round 3. Comparisons of the unsubstantiated reports at round 1 with information recorded from cards seen at round 3 indicate that mothers' reports are of very high quality; mothers' reports at round 1 were confirmed by card data at round 3 for between 83 and 93%, depending on vaccine, of children aged 12-23 months, and for 88 to 98% of children aged 24-35 months. Mothers of children who had not been vaccinated were more likely to give consistent responses than were mothers of vaccinated children. Thus, these "card plus history" estimates slightly understate true coverage levels. Most of the inconsistencies between round 1 and round 3 data apparently arose from interviewer or data processing error rather than from misreporting by mothers.