Partial tracking plays an important role in sinusoidal modeling analysis, being the stage in which the model parameters are obtained. This is accomplished by coherently grouping the spectral peaks found in each frame into time-evolving tracks of varying frequency and amplitude. The main difficulties faced by partial tracking algorithms are the analysis of polyphonic signals and the pursuit of tracks exhibiting strong modulations in frequency and amplitude. In these circumstances, linear prediction over the trajectory of a given track has been shown to improve partial tracking performance. This paper proposes an adaptive RLS lattice filter for the purpose of prediction in partial tracking. A new heuristic which certifies the filter convergence is also presented. Computer simulation results are shown to compare the proposed implementation with that of other predictors. The performance of the proposed solution is similar to that of competing methods, albeit with reduced computational complexity as well as improved numerical stability. 1