This study explores various scenarios and flexibility mechanisms to integrate high penetrations of renewable energy into the US (United States) power grid. A linear programming model - POWER (Power system Optimization With diverse Energy Resources) - is constructed and used to (1) quantify flexibility cost-benefits of geographic aggregation, renewable overgeneration, storage, and flexible electric vehicle charging, and (2) compare pathways to a fully renewable electricity system. Geographic aggregation provides the largest flexibility benefit with similar to 5-50% cost savings, but each region's contribution to the aggregate RPS (renewable portfolio standard) target is disproportionate, suggesting the need for regional-and-resource-specific RPS targets. Electric vehicle charging yields a lower levelized system cost, revealing the benefits of demand-side flexibility. However, existing demand response price structures may need adjustment to encourage optimal flexible load in highly renewable systems. Two scenarios with RPS targets from 20% to 100% for the US (peak load similar to 729 GW) and California (peak load similar to 62 GW) find each RPS target feasible from a planning perspective, but with 2x the cost and 3x the over generation at a 100% versus 80% RPS target. Emission reduction cost savings for the aggregated US system with an 80% versus 20% RPS target are roughly $200 billion/year, outweighing the $80 billion/year cost for the same RPS range. (C) 2016 Elsevier Ltd. All rights reserved.