In the last few years, musicians have been exploring ways to perform with people in different parts of a city, a country or the world. New technologies for the Internet have been developed that are already high-quality in terms of audio experience. There is, however, a persistent problem: telecommunications delay. Even though these latencies are presently down to approximately 150 milliseconds from opposites sites of the globe, we know that delays of 20 milliseconds are already problematic for musical performance. This research presents an experimental analysis of the rhythmic strategies that humans use to stay synchronized with different delay conditions, similar to the ones encountered in present day networked music performance concerts. The vast majority of experimental studies of sensorimotor synchronization involve humans synchronizing to machines, usually some sort of tapping experiments. Likewise, rhythmic tracking has overwhelmingly focused on the analysis of one musical source, but not on the interaction of performing musicians. This research's focus is on the latter. The results show that performers are able to adapt to different delay conditions anticipating each other's beats. This anticipation is believed to be an intrinsic aspect of beat perception. The results are compared to earlier studies and new metrics are proposed that consider the alternating interaction between musicians. This approach also adopts phase dynamics using stroboscopic mapping. An integrated model for tracking and generating rhythmic interactions between two performers is also presented. This model uses coupled adaptive oscillators and includes an anticipation and reaction parameter that is shown to be critical to understand rhythmic synchronization with delay. We show how this model compares to the experimental data and use it to explain some common observations in rhythmic performance with delay.