INTRODUCTION During the extension of an access road on the Canadian Forces Base (CFB) Esquimalt in Colwood, British Columbia, a black, viscous, liquid hydrocarbon product was observed oozing from newly exposed bedrock fractures in the roadside. Road excavation was subsequently halted to undertake assessment and remediation of the hydrocarbon product. The exposed bedrock was dammed with sawdust, a geomembrane barrier was installed and the area was backfilled until an appropriate course of action could be determined. The site location is shown in Figure 1. The property boundary and key site features are shown in Figure 2; the hydrocarbon seep is shown in Figure 3. Bunker C oil is a heavy-end (high molecular weight) hydrocarbon product that has a specific gravity slightly less than water and is therefore a light non-aqueous phase liquid (LNAPL). The source of the LNAPL was inferred to originate from a decommissioned fuel depot located approximately 100 m distance uphill from the road, where 40,000 barrels of Bunker C fuel oil were historically stored in one of three, large above-ground storage tanks (ASTs). A Bunker C oil spill reportedly took place at the tank farm more than two decades prior; however, the spill volume was unknown and initial investigations found no evidence of contamination between the roadside LNAPL occurrence and the former AST. Furthermore, there was also anecdotal evidence that an historic asphalt manufacturing facility may have operated in the vicinity of the access road. A hydrocarbon product similar to Bunker C is used in the manufacture of asphalt. The source and extent of the LNAPL and the potential migration pathways to the roadside location were therefore unknown. When an LNAPL spill occurs in the subsurface, the LNAPL can migrate downward under gravity through the soil pore space in the unsaturated zone. When LNAPL encounters the ground water table (the top of the saturated zone), it tends to spread out laterally because it is less dense than water and will migrate primarily in the direction of the water table gradient (water table slope). However, when LNAPL encounters bedrock, the direction of LNAPL migration can become much more complicated depending on the degree and orientation of bedrock fractures that control its movement. When fracture density is sufficiently high and the fractures are interconnected, contamination is able to migrate down-gradient through the bedrock in the same manner as through unconsolidated materials. However, when dominant structural features are prevalent that favour specific orientations, preferential pathways are created that can result in the cross-gradient migration of LNAPL. This paper presents a case study for the assessment and remediation of LNAPL in bedrock at the Canadian Forces Base (CFB) Esquimalt (the site). Geological mapping of bedrock structural features has long been used by the mining industry to identify key structures associated with economic zones of mineralization and to predict the location and extent of mineralized targets. In a similar regard, to effectively remediate LNAPL within fractured bedrock requires the identification and characterization of any structural features that might be controlling the preferential migration of LNAPL within the subsurface to other areas of the site. A significant amount of surface outcrop is present at the site and this was recognized as a cost-effective opportunity to complete a geological assessment of the bedrock. A geological mapping program was subsequently undertaken to assess bedrock outcrops for fracture density, fracture aperture, the orientation of primary fracture sets and lithologic contacts. The area was also inspected for larger scale structural features such as faults, deformation and erosional features that might influence contaminant migration. Fracture sets and lithologic contacts were mapped by outcrop location, and fractures with visible LNAPL were mapped separately from those without LNAPL. The geological data collected was used to construct stereographic projections of structural planes on a stereonet. Poles to structural planes were plotted and colour-coded by area and by presence/absence of LNAPL. The plots were then analyzed individually, and as a composite plot, to identify the dominant preferential pathways controlling LNAPL migration at the site. By superimposing these features on areas where LNAPL was observed, LNAPL delineation targets were effectively identified and the plume was subsequently delineated with confidence and remediated.