Seizure risk has consistently been found to be higher in offspring of mothers than of fathers with epilepsy. This pattern cannot be explained by any simple genetic model. The present study examined the possibility that the pattern arises from differences between affected mothers and fathers in the characteristics of their epilepsy that influence offspring seizure risk. The study population comprised 687 offspring of parents with epilepsy from the Rochester-Olmsted County Record Linkage Project. Cumulative incidences of unprovoked seizures to age 25 were 8.7% and 2.4% in offspring of affected mothers and fathers, respectively. Cox proportional hazards analysis was used to calculate rate ratios (RRs) for unprovoked seizures in offspring. In the univariate analysis, risk of unprovoked seizures was higher if the affected parent was the mother (RR = 2.8, 95% confidence interval [ci] 1.1-7.2) or if the parent's onset was before age 20 (RR = 2.5, 95% ci 1.1-5.9), but there was no effect on offspring risk of either parent's etiology (idiopathic vs. remote symptomatic) or parent's seizure type (generalized vs. partial). These findings were not substantially changed in the multivariate analysis. Thus, differences between affected mothers and fathers in these characteristics did not account for the higher risk in offspring of affected mothers. Anticonvulsant use during pregnancy was not associated with increased offspring seizure risk.(ABSTRACT TRUNCATED AT 250 WORDS)