This study proposes an intelligent PEV charging scheme that significantly reduces power system cost while maintaining reliability compared to the widely discussed valley-fill method of aggregated charging in the early morning. This study considers optimal PEV integration into the New York Independent System Operator's (NYISO) day-ahead and real-time wholesale energy markets for 21 days in June, July, and August of 2006, a record-setting summer for peak load. NYISO market and load data is used to develop a statistical Locational Marginal Price (LMP) and wholesale energy cost model. This model considers the high cost of ramping generators at peak-load and the traditional cost of steady-state operation, resulting in a framework with two competing cost objectives. Results show that intelligent charging assigns roughly 80% of PEV load to valley hours to take advantage of low steady-state cost, while placing the remaining 20% equally at shoulder and peak hours to reduce ramping cost. Compared to unregulated PEV charging, intelligent charging reduces system cost by 5-16%; a 4-9% improvement over the flat valley-fill approach. Moreover, a Charge Flexibility Constraint (CFC), independent of market modeling, is constructed from a vehicle-at-home profile and the mixture of Level 1 and Level 2 charging infrastructure. The CFC is found to severely restrict the ability to charge vehicles during the morning load valley. This study further shows that adding more Level 2 chargers without regulating PEV charging will significantly increase wholesale energy cost. Utilizing the proposed intelligent PEV charging method, there is a noticeable reduction in system cost if the penetration of Level 2 chargers is increased from 70/30 to 50/50 (Level 1/Level 2). However, the system benefit is drastically diminished for higher penetrations of Level 2 chargers. (C) 2011 Elsevier B.V. All rights reserved.