Abstract. The Agulhas Current Time-series mooring array (ACT) measured transport of the Agulhas Current at 34° S for a period of 3 years. Using along-track satellite altimetry data directly above the array, a proxy of Agulhas Current transport was developed based on the relationship between cross-current sea surface height (SSH) gradients and the measured transports. In this study, the robustness of the proxy is tested within a numerical modelling framework, using a 34-year long regional-hindcast simulation from the Hybrid Coordinate Ocean Model (HYCOM). Two reference proxies were created using HYCOM data from 2010–2013, extracting model data at the mooring positions and along the satellite altimeter track for; (1) the box transport (Tbox) and (2) the jet (southwestward) transport (Tjet). Next, sensitivity tests were performed where the proxy was recalculated from HYCOM for (1) a period where the modelled vertical stratification was different compared to the reference proxy, and (2) different lengths of periods: 1, 3, 6, 12, 18 and 34 years. Compared to the simulated (native) transports, it was found that the HYCOM proxy was more capable of estimating the box transport of the Agulhas Current compared to the jet transport. The HYCOM configuration in this study contained exaggerated levels of offshore variability in the form of frequently-impinging baroclinic anticyclonic eddies. These eddies consequently broke down the linear relationship between SSH slope and vertically-integrated transport, resulting in stronger correlations for the inshore linear regression models compared to the ones offshore. Vertically-integrated transport estimates were therefore more accurate inshore than those offshore or when the current was in a meandering state. Results showed that calculating the proxy over shorter or longer time periods in the model did not significantly impact the skill of the Agulhas transport proxy, suggesting that 3-years was a sufficiently long time-period for the observation based transport proxy.