Objective Previous work has suggested that seizure outcome is the most important predictor of quality of life (QoL) after epilepsy surgery, but it is unknown which specific seizure outcome measure should be used in judging surgical success. We assess three different seizure outcome measures (relative seizure reduction, absolute seizure reduction, and seizure freedom [yes/no]) to investigate which measure best predicts postoperative QoL. Methods We prospectively surveyed patients at outpatient visits before and after epilepsy surgery (n = 550). The QoL measure was the Quality of Life in Epilepsy (QOLIE‐10) score at the patient's most recent office visit. We created multivariate regression models to predict postoperative QOLIE‐10, with a different seizure outcome measure in each model. We compared models using adjusted R2 values and Akaike information criteria (AIC). Results Our cohort had a high level of disease severity and complexity (17% repeat surgery, 39% extratemporal, and 18% nonlesional). For the cohort as a whole, mean absolute seizure frequency decreased from 1 per day to 0.1 per day (P < .001), and mean reduction was 73% (95% confidence interval [CI] 66%‐81%). Average improvement in QoL score was 5.3 (95% CI 4.1‐6.5) points. Of patients who reported an improvement in QoL, 27% had persistent seizures. Comparison of regression models to predict QoL showed that the worst model was provided when using “absolute seizure reduction,” but that models using “relative seizure reduction” and “seizure freedom (yes/no)” were equally strong. Significance In our high severity and complexity cohort, a substantial subset of patients (27%) reported improved QoL despite persistent seizures. Relative seizure reduction was at least as good a predictor of QoL as seizure freedom. A yes/no seizure freedom variable may be a suboptimal measure of surgical success, especially in high complexity cohorts.