We compared rotavirus detection rates in children with acute gastroenteritis (AGE) and in healthy controls using enzyme immunoassays (EIAs) and semiquantitative real-time reverse transcription PCR (qRT-PCR). We calculated rotavirus vaccine effectiveness using different laboratory-based case definitions to determine which best identified the proportion of disease that was vaccine preventable. Of 648 AGE patients, 158 (24%) were EIA positive, and 157 were also qRT-PCR positive. An additional 65 (10%) were qRT-PCR positive but EIA negative. Of 500 healthy controls, 1 was EIA positive and 24 (5%) were qRT-PCR positive. Rotavirus vaccine was highly effective (84% [95% CI 71%-91%]) in EIA-positive children but offered no significant protection (14% [95% CI -105% to 64%]) in EIA-negative children for whom virus was detected by qRT-PCR alone. Children with rotavirus detected by qRT-PCR but not by EIA were not protected by vaccination, suggesting that rotavirus detected by qRT-PCR alone might not be causally associated with AGE in all patients.