Identification of surface hydrological connectivity at scales where critical source areas (CSAs) can be managed is fundamental to achieving effective management of phosphorus loss in agricultural catchments. This study investigated the potential for the 'Network Index' (NI) to predict surface connectivity at CSA-management scales in contrasting agricultural catchments (ca. 1200 ha) using a 5 m DEM. One catchment had mostly permeable soils and a low surface ditch density. The other catchment had mostly impermeable soils and a high surface ditch density. The importance of surface ditch data for accurately modelling the location and magnitude of surface connectivity was also evaluated. Modelled surface connectivity at the subcatchment scale (ca. 130 ha) was validated using observed channel (ditch and stream) densities. Modelled surface connectivity at the field scale (ca. 2 ha) was evaluated using four indicators of observed field connectivity. At the subcatchment scale, modelled surface connectivity matched observations well (R^2 = 0.52) despite the soil type variability across catchments. However, some errors in subcatchment boundary delineation (total of 133 ha in one catchment) occurred which could distort the extent of predicted CSAs. At the field scale, the NI had potential for broadly discerning the most connected from the least connected fields which is valuable for identifying where CSA-based management should be targeted. Detailed ditch information was required to accurately delineate subcatchment boundaries; however, it was not needed for predicting subcatchment connectivity.