This thesis has examined the 30km long rail corridor through the Lower Buller Gorge, on the Stillwater Ngakawau Line, between SNL96 and 126km, using a landslide risk management approach. The project area is characterised by high annual rainfall (>2,000mm per year), and steep topography (slopes typically ≥20°) adjacent to the rail corridor. The track formation generally follows the natural contour near the base of the hillslope through the Lower Buller Gorge, and consequently involves many curves but relatively limited cut slopes into adjacent rock outcrops. The distance between the base of adjacent hillslopes and rail is frequently <2m horizontally. A variety of basement and Tertiary lithologies are present, including granite, breccias, indurated sandstone/mudstone, and limestone. The primary focus of this thesis has been on upslope-sourced landsliding onto the rail corridor, and on two short lengths (20m and 450m) that currently have a 25km/hour speed restriction imposed at Whitecliffs and Te Kuha respectively. Rainfall-induced and earthquake-generated landslide triggering mechanisms were examined in detail. A landslide inventory has been compiled to determine the characteristics and distribution of identified slope failures over time, and to establish any correlation with topography and geology. Sixty individual landslide events were identified since the line became fully operational in the 1940s, based on desktop reviews, and field inspections for more recent events. To reflect the presence of small magnitude landslide events, a project-specific logarithmic classification of landslides was adopted from <10m³ (very small volume) to ≥10,000m³ (very large volume). An absence of a higher proportion of ‘very small’ to ‘small’ landslide volumes (<100m³) in the inventory reflects incomplete reporting of these comparatively lower magnitude, but higher frequency, events. The establishment of a robust landslide inventory to document future events, in a consistent and readily accessible format, is required for continued monitoring and review of landslide risk management practices in the Lower Buller Gorge. Combining landslide inventory data and physical characteristics of the project area enabled the development of a qualitative landslide zonation map that assigned ‘high’, ‘high-moderate’, ‘moderate’ and ‘low’ landslide susceptibility classes. The principal area of slope instability above the rail corridor is 22.5km in length between SNL103.5 and 126.0km, associated predominantly with basement lithologies (Tuhua Granite; Hawks Crag Breccia; Greenland Group). The most frequently occurring landslides are shallow, typically less than 3m deep, translational failures triggered in regolith or colluvium materials. Rainfall-induced debris slides and flows are dominant, given the high annual rainfall and associated high frequency of high intensity or long duration rainfall events. Very small to medium landslides (<1,000m³) have the potential to impact the rail corridor with an average frequency of around one every two years, causing damage to infrastructure or affecting rail operations. Very large landslides (≥10,000m³) can be expected every 10 to 20 years based on a limited historical record. The narrow rail corridor and absence of sufficient catch areas above or adjacent to the rail causes continual operational challenges due to upslope-sourced landslide debris, and high susceptibility to slope failures, particularly west of SNL103.50km. Development of a rainfall-threshold for proactive inspection of the rail corridor is recommended, including the establishment of a rain gauge network through the Lower Buller Gorge. Earthquake-generated landslides significantly impacted the rail during the magnitude 7.1 Inangahua earthquake in 1968 and to a much lesser extent during the magnitude 6.1 Westport earthquake in 1991. The rail was not fully constructed through the Lower Buller Gorge at the time of the magnitude 7.8 Buller (Murchison) Earthquake in 1929, which generated widespread landsliding in the Buller and Nelson regions. Earthquake-generated landsliding can be expected through the Lower Buller Gorge from earthquakes of magnitude ≥6, and track inspection is recommended in the event of magnitude 5 or greater earthquakes. Detailed geological characterisation and mapping at Whitecliffs and Te Kuha was conducted, including a LiDAR survey at Whitecliffs that enabled visualisation of the ground surface without the interference of vegetation. The limestone outcrop at Whitecliffs comprises 60-70m high near-vertical cliffs with a well-established talus apron at the base, extending to the rail corridor. Three widely spaced open fractures sets are present at the top of Whitecliffs that propagate into the cliff-face. There has been no detectable movement on selected key fracture sets since monitoring commenced in 1993 and there is no confirmed evidence of large-scale cliff collapse during the 1968 Inangahua earthquake. Whitecliffs is not as susceptible to failure as other slopes inspected in the project area due to structural controls, primarily being the dipping of strata back into the cliff-face and widely space joint sets. Establishment of inspection protocols for earthquake events impacting the area, including real-time monitoring of selected fractures at Whitecliffs is recommended. A 2km-length corridor site model produced for Te Kuha demonstrated ‘high’ landslide susceptibility is not confined to slopes above the existing 450m speed restriction zone. Removal of the speed restrictions at Whitecliffs and Te Kuha can be considered, as the increased exposure time is not considered sufficient justification given the extent of other susceptible areas to landsliding affecting the Lower Buller Gorge rail corridor. The principal conclusion from this thesis project is that there is on-going risk to rail operations predominantly from shallow translational landsliding in regolith-colluvium materials. The majority of these will be generated by long-duration or intense rainfall events. Development of threshold-based methods for effective track management is recommended, including the establishment of a rain gauge network through the Lower Buller Gorge, and landslide inventory database. Site-specific engineering measures could be adopted, such as catch benches or avalanche-type shelters, where justified on a cost-benefit basis.