This paper describes an optimization procedure to modify the geometry of a mixed-flow turbocharger turbine for improved aerodynamic efficiency. The procedure integrates parameterization of the turbine blade geometry, genetic algorithm optimization, and 3D CFD analysis using a commercial solver. Using a known mixed-flow turbocharger turbine as the baseline, the main features of the blade geometry-the hub, shroud, camberline, leading and trailing edge profiles-were separately adjusted by the genetic algorithm in the direction of better efficiency. Apart from optimizing the subject turbine for the operating point in question, more usefully this permits each geometrical feature to be ranked by their contribution to the change in efficiency. Cases were also run in which the hub and shroud curves were simultaneously adjusted. Analysis of CFD results provides additional insight into the underlying reasons for efficiency changes by examination of the relevant flow field features. The hub and shroud profiles were observed to have the greatest impact on turbine performance, optimization of which leads to an increase of 1.3 percentage points of efficiency. This compares to only 0.2 percentage points improvement following optimization of the outlet geometry.