About one-third of people with epilepsy remain drug-resistant, and surgical outcomes vary. We investigated whether sleep-resolved normative mapping of intracranial EEG (iEEG) – separating periodic bandpower from aperiodic activity – can more reliably localize epileptogenic tissue. Using non-seizure-onset-zone (non-SOZ) electrodes, we built sleep-stage (Wake, NREM, REM) normative atlases and computed age- and sex-adjusted W-scores. Features included relative band power (δ, θ, α, β, γ) and aperiodic exponent and offset. We compared SOZ versus non-SOZ electrodes, quantified univariate discrimination with ROC AUCs, and trained multivariate models with out-of-fold evaluation to assess stage-specific classification. Qualitatively, spatial maps showed clear, stage-dependent shifts in spectral metrics. Across stages, the aperiodic offset consistently distinguished SOZ from non-SOZ, whereas the aperiodic exponent contributed little. Several temporal-lobe regions yielded robust, stage-specific univariate AUCs, with left-lateralized SOZs generally more separable. Lastly, multivariate models maintained strong SOZ-non-SOZ discrimination across sleep, and aperiodic offset repeatedly ranked among the top features. Together, these findings indicate normative iEEG mapping via sleep staging with W-scores identifies epileptogenic tissue by leveraging complementary periodic and aperiodic signatures. Aperiodic offset emerges as a stable, high-value marker, while stage-specific periodic features add regional nuance. This work supports integrating sleep-resolved normative atlases into presurgical evaluation for SOZ resection, and motivates future work incorporating white-matter-tract regionalization to improve lateralization and target selection in drug-resistant epilepsy patients.