Despite advances in intracranial EEG (iEEG) technique, technology and neuroimaging, patients today are no more likely to achieve seizure freedom after epilepsy surgery than they were 20 years ago. These poor outcomes are in part due to the difficulty and subjectivity associated with interpreting iEEG recordings, and have led to widespread interest in developing quantitative methods to localize the epileptogenic zone. Approaches to computational iEEG analysis vary widely, spanning studies of both seizures and interictal periods, and encompassing a range of techniques including electrographic signal analysis and graph theory. However, many current methods often fail to generalize to new data and are sensitive to differences in pathology and electrode placement. Indeed, none have completed prospective clinical trials. In this dissertation, I develop and validate tools for guiding epilepsy surgery through the quantitative analysis of intracranial EEG. Specifically, I leverage methods from graph theory for mapping network synchronizability to predict surgical outcome from ictal recordings, and also investigate the effects of sampling bias on network models. Finally, I construct a normative intracranial EEG atlas as a framework for objectively identifying patterns of abnormal neural activity and connectivity. Overall, the methods and results of this dissertation support the implementation of quantitative iEEG analysis in epilepsy surgical evaluation.