During the monitoring of pharmacoresistant epilepsy patients prior to surgery, interictal epileptic discharges (IEDs) are analyzed to locate possible sources of epileptic activity. In order to compensate low spatial resolution of EEG, simultaneous EEG-fMRI recordings can be used. Conventional methods typically deploy an EEG-informed analysis of the fMRI data; i.e., EEG-derived IED onset timings are used to setup regressors for linear regression. Recently we have proposed a new fMRI analysis method, total activation (TA), which is able to deconvolve the underlying activity-inducing signal without prior information on the onset timing and duration of events. Here we demonstrate that TA can locate the epileptogenic regions from fMRI data. We compare and validate our results with conventional methods performed by experts prior to surgery.