This chapter presents a comprehensive analytical framework for characterizing the charging demand of plug-in electric vehicles (PEVs) under large-scale penetration scenarios, addressing the critical need for accurate load forecasting to ensure the stable and secure operation of power grids. As PEV adoption increases, the associated charging load introduces significant uncertainties and potential risks to grid operations, necessitating a deep understanding of the spatiotemporal patterns of PEV charging behavior. To tackle the challenge of limited real-world data due to low current PEV penetration, this chapter proposes three complementary modeling approaches. First, a quantitative analysis method based on time and electricity consumption is developed, utilizing Monte Carlo simulation to generate large-scale synthetic travel and charging data grounded in existing travel surveys and battery parameters. Second, a Markov model is introduced to capture the state transition probabilities of PEVs over time, enabling the prediction of vehicle state distributions, e.g., driving, parking, and charging—across different time intervals throughout the day. Third, a hidden Markov model (HMM) incorporating battery state of charge (SoC) is proposed to reveal the probabilistic relationship between vehicle states and battery energy levels, allowing for the inference of SoC distributions from observable state sequences. These models collectively enable accurate prediction of PEV charging demand and energy status over time. Furthermore, the chapter investigates the adverse impacts of uncoordinated PEV charging on the power grid, including increased peak-valley differences and voltage drops, using real grid load data and IEEE benchmark systems. The findings underscore the importance of developing scheduling mechanisms and infrastructure plans to mitigate these effects, providing essential insights for grid operators and policymakers.