Head Checking (HC) is a major type of Rolling Contact Fatigue (RCF) in railway rails across the globe. It mainly occurs on curved tracks in the rail shoulder of the gauge side and at the gauge corner because of the large lateral force. The related track radii are between 500 – 3000 m. It initiates from the surface due to high surface shear stresses arising at wheel-rail contact. HC has severe economic consequences as well as on the safety of railway operations. The serious accident caused by HC at Hatfield in the United Kingdom in October 2000 raised awareness to treat it seriously. The yearly total HC treatment-related cost was about 50 million euros in the Netherlands when the occurrence of HC was at its highest. Although a number of treatment methods for HC are possible, it is concluded that preventing or retarding HC initiation by optimal rail profile design is the most effective in terms of implementability, cost and time span. This thesis therefore aims at the design of an anti-HC profile of rails, based on a fundamental understanding of the mechanical mechanism of HC initiation. To such end, an investigation has been carried out on the quantitative relationship between HC occurrences, contact geometry, stresses and microslip. HC initiation has been reproduced under controlled laboratory conditions on a full-scale wheel-rail test rig. At the same time, HC initiation has been monitored in the field under service conditions. Using a non-Hertzian rolling contact solution method, it is found that HC initiation location tends to be at a distance 7 – 12 mm from the gauge face, where the surface shear stress is the highest as a result of the large geometrical spin in the wheel-rail contact. The optimization is therefore focused on the gauge part of the profile, with the objective of relieving the maximum shear stress. As the 54E1 rail is predominantly used on the Dutch railway network, the optimization is performed on it. After a statistical analysis of the AHC performance of the 54E1 and 46E3 profiles, it is concluded that an undercut of the 54E1 profile at the gauge corner, with the maximum undercut at about 9 mm from the gauge face, should achieve the objective. Together with a number of constraints arising from the existing 54E1 profile, from vehicle running performance, track structure and contact mechanics, an optimal Anti Head Checking 54E1 (AHC 54E1) profile is designed. This designed profile has shown its merits: By avoiding contact in the HC-prone part of the rail, the maximum surface shear stress is greatly reduced, mainly owing to the decrease of spin in the contact. A monitored field test shows that the AHC 54E1 profile can largely delay the HC formation and once HC arises, it also decreases the crack growth by a factor of half. The AHC profile changes due to wear, so that it has to be restored with cyclic grinding to maintain its effectiveness. Large-scale application on the Dutch railway network shows that HC in 2008 was reduced by about 70% with respect to 2004 when HC was the most widespread. At the same time, no negative influence of the AHC 54E1 on the running performance of the trains has been reported, either from the monitored site or from the large-scale application. As a result, the AHC 54E1 profile has been normalized as a standard European rail profile named 54E5 at 1:40, see prEN 13674-1, June, 2009. Recommendations for further research and development are made at the end of the thesis.