A new constraint assignment algorithm and a two-perspective tracking method are presented for threedimensional tracking of motile aquatic organisms. The method aims at providing preferably long tracks of multiple, simultaneously swimming individuals. As an extension to existing tracking and assignment methods, the presented algorithm takes background knowledge about maximum swimming speed and size of the organisms into account. Two strategies, both using the constraint assignment algorithm, were applied to track either a fixed number of objects or to track as many objects as possible in frame-to-frame transitions. Whereas the first strategy provides continuous and distinct tracks of all individuals for the entire measurement sequence, the second strategy does not require interpolation of unresolved positions and is more robust to disappearance or reappearance of organisms within the field of view. Two sets of video records of freely swimming Daphnia magna, differing in abundance, image quality, and record length, are analyzed to assess the tracking success of the proposed algorithm. Although both strategies provide better results for a higher detection rate of organisms, the first strategy is preferable for tracking individuals in tests with smaller organism abundance and higher signalto-noise ratio, whereas the second strategy also provides reasonable pathways under less optimal conditions. Spectral analysis of observed swimming velocities is applied to demonstrate the advantage of obtaining long, continuous, and three-dimensional tracks of moving organisms.