Agroforestry can offer carbon sequestration, higher system productivity and biodiversity. However, a limited number of field trials exist to study their feasibility and trade-offs for large scale deployment. Agroecosystem models could represent a valuable tool for their ex ante assessment. Here, we present ZonalCent, a novel approach to use the DayCent model to simulate multi-component agroforestry systems by splitting them into several independent zones, and simulating each zone individually. We used six agroforestry sites in France to evaluate how well ZonalCent represented carbon sequestration in tree biomass, soil organic carbon stocks and in the total system. This proved promising because with the default parameter set of DayCent, ZonalCent was highly suitable to represent tree carbon sequestration (Nash-Sutcliffe modelling efficiency; NSE of 0.82), and for total system carbon sequestration potential (NSE of 0.67), despite a tendency to overestimate SOC stocks (NSE of 0.31). Only one site had yield data and there, ZonalCent approach could approximate the mean yield reduction - yet more detailed evaluation is necessary. Negative correlations with yield showed that yield was most strongly affected by a) shading by mature trees and b) the loss of arable area due to grass strips. While more detailed models may be needed for a detailed process understanding, ZonalCent includes the most important interactions (light, water, nutrients, temperature) in a simple but effective way and can be readily used - because it is based on DayCent - to estimate the potential carbon sequestration of agroforestry systems at larger scales.