Background Studies have reported a temporal association between a first dose of rotavirus vaccine (Rotashield) and infant intussusception. We investigated the effect of Rotashield vaccination use on intussusception admissions in ten US states. Methods We analysed electronic databases containing 100% hospital discharge records for 1993-99 from ten US states, where an estimated 28% of the birth cohort had received Rotashield (based on manufacturer's net sales data). We examined records of infants admitted to hospital (<365 days old) with any mention of intussusception (International Classification of Diseases, ninth revision, clinical modification code 560 times 0). Excess admissions for intussusception during the period of Rotashield availability (October 1998 to June 1999) were estimated by direct comparison with the corresponding period of October 1997 to June 1998 (before Rotashield was available) and with adjustment for secular trends during 1993-98 by Poisson regression. Findings Hospital admission for intussusception among infants younger than 365 days of age during the Rotashield period compared with previously was 4% lower (10 cases) by direct comparison and 10% lower (27 cases) by trend comparison, corresponding to a negative population attributable risk. Among infants aged 45-210 days (target age range for a first Rotashield dose), we estimated an increase in intussusception admissions of 1% (one excess admission) by direct comparison and 4% (4 times 6 excess admissions) by trend comparison, corresponding to a population attributable risk range of one excess admission in 66 000-302 Interpretation We found no evidence of increased infant intussusception admissions during the period of Rotashield availability. The total intussusception admission risk attributable to Rotashield was substantially lower than previous estimates based on studies focusing on the immediate postvaccination weeks.