We investigate the skill of the Community Earth System Model version 1 (CESM1) in simulating hydrologic processes over the Mississippi River basin. Simulated discharge is seasonally delayed relative to observations: model biases are diagnosed using variables from reanalysis data and attributed primarily to precipitation and runoff related processes. We also show that the seasonality of simulated runoff in several Coupled Model Intercomparison Phase 6 (CMIP6) models is improved relative to CESM1.