The paper presents some improvements to a novel approach for the determination of 3-D flow velocity fields, which is based on 3-D particle tracking in a tomographic reconstruction of an observation volume seeded with tracer particles. The basic idea of the approach is using multiple camera views for a full tomographic reconstruction of the object space, which is represented by a 3-D voxel structure with a resolution adapted to the camera resolution. Based on the images of four or more convergent cameras with their orientation known from a prior calibration procedure, a complete 3-D light intensity distribution in the observation volume can be reconstructed by a projective transformation of each camera image contents into each depth layer of the object space and a consecutive minimum search. 3-D velocity field information can then be obtained by volume-based tracking in time-resolved voxel space representations. This procedure represents a rather elegant way of completely avoiding detection and matching ambiguities, thus allowing for a significant increase of the spatial resolution of 3-D particle tracking. The paper will show the basic concept of tomographic reconstruction and tracking in 3D-PTV and show some first results from processing synthetic data sets. The computational effort, accuracy and spatial resolution potential of the technique will be compared to conventional 3-D particle tracking velocimetry.