The recent advances in the chronic implantation of electrodes have allowed the collection of extracellular activity from neurons over long periods of time. To fully take advantage of these recordings, it is necessary to track single neurons continuously, particularly when their associated waveform changes with time. Multiple spike sorting algorithms can track drifting neurons but they do not perform well in conditions like a temporary increase in the noise level, sparsely firing neurons, and changes in the number of detectable neurons. In this work, we present Spikes_Link, a general framework to track neurons under these conditions. Spikes_Link can be implemented with different spike sorting algorithms, allowing the experimenter to use the algorithm best fitted to their recording setup. The main idea behind Spikes_Link is the blockwise analysis of the recording using overlapping sets of spikes to equally represent all the putative neurons being tracked on a given block. This way, we can link classes with clusters obtained in a new block based on an overlapping metric. Moreover, the algorithm can fix temporary sorting errors (splits and merges). We compared an implementation of Spikes_Link with other algorithms using long-term simulations and obtained superior performance in all the metrics. In general, the Spikes_Link framework could be used for other clustering problems with concept drift and class imbalance.