Prior to the detailed design of components, turbomachinery engineers must guide a mean-line or throughflow design toward an optimum configuration. This process requires a combination of informed judgement and low-order correlations for the principle sources of loss. With these requirements in mind, this paper examines the impact of key design parameters on endwall loss in turbines, a problem which remains poorly understood. This paper presents a parametric study of linear cascades, which represent a simplified model of real-engine flow. The designs are nominally representative of the low-pressure turbine blades of an aero-engine, with varying flow angles, blade thickness, and suction surface lift styles. Reynolds-averaged Navier–Stokes (RANS) calculations are performed for a single aspect ratio (AR) and constant inlet boundary layer thickness. To characterize the cascades studied, the two-dimensional design space is examined before studying endwall losses in detail. It is demonstrated that endwall loss can be decomposed into two components: one due to the dissipation associated with the endwall boundary layer and another induced by the secondary flows. This secondary-flow-induced loss is found to scale with a measure of streamwise vorticity predicted by classical secondary flow theory.