The production potential of switchgrass (Panicum virgatum L.) has not been estimated in a Mediterranean climate on a regional basis and its economic and environmental contribution as a biofuel crop remains unknown. The objectives of the study were to calibrate and validate a biogeochemical model, DAYCENT, and to predict the biomass yield potential of switchgrass across the Central Valley of California. Six common cultivars were calibrated using published data across the US and validated with data generated from four field trials in California (2007–2009). After calibration, the modeled range of yields across the cultivars and various management practices in the US (excluding California) was 2.4–41.2 Mg ha−1 yr−1, generally compatible with the observed yield range of 1.3–33.7 Mg ha−1 yr−1. Overall, the model was successfully validated in California; the model explained 66–90% of observed yield variation in 2007–2009. The range of modeled yields was 2.0–41.4 Mg ha−1 yr−1, which corresponded to the observed range of 1.3–41.1 Mg ha−1 yr−1. The response to N fertilizer and harvest frequency on yields were also reasonably validated. The model estimated that Alamo (21–23 Mg ha−1 yr−1) and Kanlow (22–24 Mg ha−1 yr−1) had greatest yield potential during the years after establishment. The effects of soil texture on modeled yields tended to be consistent for all cultivars, but there were distinct climatic (e.g., annual mean maximum temperature) controls among the cultivars. Our modeled results suggest that early stand maintenance of irrigated switchgrass is strongly dependent on available soil N; estimated yields increased by 1.6–5.5 Mg ha−1 yr−1 when residual soil mineral N was sufficient for optimal re‐growth. Therefore, management options of switchgrass for regional biomass production should be ecotype‐specific and ensure available soil N maintenance.