Abstract Cross-correlation Particle Image Velocimetry (PIV) has become a well known and widely used experimental technique. It has been already documented that difficulties arise resolving velocity structures smaller than the interrogation window. This is caused by signal averaging over this window. A new cross-correlation PIV method that eliminates this restriction is presented. The new method brings some other enhancements, such as the ability to deal with large velocity gradients, seeding density inhomogeneities, and high dispersion in the brightness of the particles. The final result is a method with a remarkable capability for accurately resolving small scale structures in the flow, down to a few times the mean distance between particles. When compared to particle tracking velocimetry, the new method is capable of obtaining measurements at high seeding density concentrations. Therefore, better overall performance is obtained, especially with the limited resolutions of video CCDs. In this paper, the new method is described and its performance is evaluated and compared to traditional PIV systems using synthetic images. Application to real PIV data are included and the results discussed.