Pipeline routes have traditionally been selected in the early stages of prospect development, using low-resolution datasets, with minimal consideration for seafloor and shallow subsurface geological conditions, and rudimentary rules about seafloor gradients and pipeline curvature. We describe advantages of a modern GIS-based approach using least-cost path optimization over a composite geocost surface, which provides a structured framework for quantifiable decision-making, incorporates formal risk assessment as part of the route acceptance process, and forms the basis for a forthcoming industry guidance document. The workflow that we propose comprises: 1) geophysical data acquisition, geotech-nical investigations, and geological studies during the initial stages of development to evaluate potential pipeline routes; 2) systematic evaluation of geological, geotechnical, ecological, and cultural factors that determine pipeline route suitability; 3) the classification and weighting of all identified geohazards and constraints to develop a geocost composite map; 4) determination of an optimal route using least-cost path optimization within a geographic information system; 5) risk assessment of the least-cost route; and, finally, 6) route acceptance. The classification and weighting of component hazard maps, as well as the risk assessment, are best performed collaboratively by a multi-disciplinary team that includes expertise in marine engineering geology, geotechnical engineering, pipeline engineering, marine ecology and archaeology, and health, safety, and environmental personnel. We will demonstrate a step-by-step application of the method, which has already been used for pipeline route determination on geologically challenging deepwater projects. If geohazards are not identified in early stages and used to guide route selection, changes to the original pipeline route or mitigation - either of which could affect schedules and budgets - may be necessary. This method is relatively new to subsea pipeline route determination, however, it offers important advantages over previous methods and can be applied starting in the earliest conceptual phases of field development, iterating if necessary, as new information becomes available.