We used the records of a statewide pediatric mortality surveillance system to determine mortality rates and causes of death in children with epilepsy. Of the 1,095 children aged 1-14 years who died in the state of Victoria during the study period 1985-1989, 93 had a history of epilepsy. Six children (6%) had primary epilepsy, and 87 (94%) had secondary epilepsy. Death was (a) directly attributable to epilepsy in 20 (22%), including 11 with sudden unexplained death, (b) not directly attributable to epilepsy in 59 (63%), and (c) of undetermined cause in 14 (15%). No classifiable death occurred as a direct result of status epilepticus. The average annual mortality rates for children with epilepsy were (a) death from all causes, 30.6 in 10,000 [95% confidence interval (CI) 19.7, 47.5], and (b) death attributable to epilepsy, 6.6 in 10,000 (95% CI 3.7, 11.8). Relative to the all-cause mortality rate in children without epilepsy, the all-cause mortality rate ratios were (a) all children with epilepsy, 13.2 (95% CI 8.5, 20.7); (b) primary epilepsy, 1.1 (95% CI 0.5, 2.6); and (c) secondary epilepsy, 49.7 (95% CI 31.7, 77.9). The mortality rate ratios for secondary epilepsy relative to primary epilepsy were (a) death from all causes, 43.5 (95% CI 19.0, 99.5); and (b) death attributable to epilepsy, 9.0 (95% CI 3.3, 24.8). Epilepsy appeared on the death certificate of only 11 of 20 (55%) children whose deaths were attributable to epilepsy. We conclude that (a) there was an increased risk of death during childhood in children with epilepsy; (b) the risk of death was greatest for children with secondary epilepsy; (c) potentially preventable, epilepsy-related deaths occurred in children with primary epilepsy; (d) sudden unexplained death accounted for at least 12% of deaths; and (e) death certification was deficient with respect to recording of epilepsy.