TLS is seeing increasing usage in survey practices as a means of quickly capturing large volumes of point data, with one such practice being the surveying of as-built features for roads. This paper aims to report the automation of the identification and extraction process of kerbs present in point clouds of roadways, which is a feature of primary importance when dealing with road scenes. This proposed process is done in multiple stages as follows. The first is to isolate the road surface and ground points using simple statistical classification. This leads to the identification of candidate points located near the extents of the surface that are determined to have a high likelihood of being sampled for the kerb features. Once these points have been identified, the orientation and direction of the kerb at the candidate points are approximated using principal component analysis on the local neighbourhood. A localised kerb profile can be then fitted to the candidate point, which is based on the derived neighbourhood properties and examination of an extracted cross-section. After the kerb profile has been determined, it is then incrementally extruded along the kerb from a candidate point to determine the path of the kerb. This allows for the automated extraction of the kerb feature from the TLS, as well as the determination various other properties of the road, such as its centreline. The outline method will be demonstrated on several practical datasets, with the results highlighting the simple and robust nature of the proposed method.