This paper presents an investigation of a range of engine realistic rim seals starting from a simple axial seal to different types of overlapping seals. The experiments were performed in a large-scale linear cascade equipped with a secondary air system capable of varying independently both the mass fraction as well as the swirl velocity of the leakage air. The experimental results were also complemented by computationally fluid dynamics (CFD) to provide better insight in the flow physics. It has been found that the key feature of the rim seals that affect their impact on overall loss generation and their ability to provide good sealing effectiveness was the location and the size of the recirculation zones within the rim seal. The requirements for good sealing and reduced spoiling effects on the main gaspath flow often led to contradictory designs. In general, the recirculation zones were found to improve sealing by reducing the effect of the pitchwise (circumferential) variation in the pressure distribution due to the blade's potential field, and thus reduce ingestion. However, at the same time the recirculation zones tend to increase the loss generation. The best compromise was found when the outer part of the seal and its interface with the rotor platform was as smooth as possible to minimize the spoiling losses, while the recirculation zones were confined to the inner part of the seal to maintain acceptable levels of sealing effectiveness. A new rim seal design, which utilizes the best attributes of the above mentioned designs was developed. Linear cascade tests showed the losses due to the leakage-mainstream interaction were reduced by 33% compared to the datum seal design. Further validation was performed by examining the new configuration using unsteady full-stage calculations under engine realistic conditions. These calculations suggest an improvement of nearly 0.2% in the stage efficiency.