Conventional trip assignment characterizes the path choice of inter-zonal travelers and loads the inter-zonal trips from a zonal centroid via its access points onto a network, while it cannot handle the path choice of intra-zonal travelers and keeps the intra-zonal trips out of the network. In addition, it may unrealistically concentrate trip loading on only one or two directions. In regional travel demand models, usually the percent of intra-zonal pedestrian trips is relatively high. Taking the Triangle regional model as an example, the percent of intra-zonal pedestrian trips varies from 9.9% to 34.5% by trip purpose and overall it is 22.4% for all trips. In this case the conventional assignment cannot load 22.4% of pedestrian trips onto the network and therefore has a trip under-loading problem, especially for the locations around zones with high percent of intra-zonal trips. The current developed multi-point assignment can load trips from a centroid via multiple access points onto the network based on fixed loading rate by direction or some sorts of loading control rules in order to diversify loading directions to a reasonable level. It has the capability to fix the under-loading problem and to handle more complex intra-zonal trip loading by properly splitting and expanding the trip demand matrix using loading rates of access points and assigning intra-zonal trips onto the network from one access point to others within one zone. The Durham-Chapel Hill-Carrboro MPO conducted a pilot study on implementing the multi-class multi-point assignment method for pedestrian trips. Seven trip purposes in the regional model were aggregated to three trip purposes/classes, i.e. home-based work/university, home-based other and non-home based. In consideration of possible different loading directions among these three purposes/classes, a multi-class multi-point assignment was employed with a developed tool in TransCAD for the implementation. In the areas of concern, such as school, university and downtown areas, more access points were added with proper loading rates/rules. The result shows significant improvements in these areas and most of intra-zonal pedestrian trips were assigned onto the network. It is also a quick response approach that doesn’t requires splitting the zones into smaller ones and re-run all modeling steps.