Objective: Epilepsy is common and carries substantial morbidity, and therefore identifying cost-effective health interventions is essential. Cost-utility analysis is a widely used method for such analyses. For this, health conditions are rated in terms of utilities, which provide a standardized score to reflect quality of life. Utilities are obtained either indirectly using quality of life questionnaires, or directly from patients or the general population. We sought to describe instruments used to estimate utilities in epilepsy populations, and how results differ according to methods used. Methods: We undertook a systematic review of studies comparing at least two instruments for obtaining utilities in epilepsy populations. MEDLINE, Embase, ScienceDirect, Cochrane Library, Google Scholar, and gray literature were searched from inception to June 2019. Mean utilities were recorded and compared for each method. Results: Of the 38 unique records initially identified, eight studies met inclusion criteria. Utilities were highest for direct "tradeoff" methods, obtained via instruments including standard gamble (0.93) and time tradeoff (0.92), compared to indirect methods, obtained via instruments including EuroQoL five-dimensional form (range = 0.72-0.86) and Health Utilities Index Mark 3 (range = 0.52-0.71). Visual analog scale (VAS), a direct "nontradeoff" instrument, provided equal or lower utilities (range = 68.0-79.8) compared to indirect instruments. Significance: Direct methods, with the important exception of VAS, may provide higher utilities than indirect methods. More studies are needed to identify the most appropriate utility instruments for epilepsy populations, and to investigate whether there is variation between utilities for different types of epilepsy and other patientand disease-specific factors.