We describe the problem of regular small telemetry losses incurred during coherency mode transitions in Cassini's telecommunication. The project did not originally plan any corrective steps for avoiding these data losses, because of 1) the disparity between the small durations of the transitions (1-2 min) and large playback capability losses (15 min) needed for bracketing the transition time spans and 2) the unpredictable content of data downlinking during the transitions. However, as the intense science data return from the tour began, it became apparent that the impact of these small losses can sometimes be significant. We provide two examples of the impact on Radar-dedicated Titan flybys. In general, the impacts are larger for high-rate data and for data acquired during a targeted flyby of Titan any other icy satellites. Although the content of data during a transition for every downlink pass is unpredictable, we are certain that some important data will be lost on downlink passes dedicated to transmit the flyby data and it does not matter what part of the data will be hit by the transitions. We collected more than 200 days of data from Cassini tour operations between June 2004 and February 2005 to analyze the distributions of the start time and duration of the transitions. We found that the occurrence of a transition can be predicted within a 5-min window, with 95 percent confidence. Given that, it is possible to eliminate the data losses by pausing playback at the beginning of a transition for 5 minutes and resuming playback after transition completion. We briefly describe three operational fixes as to how to implement the playback pause, with the pros and cons for each method. Finally, we report the results of the method chosen by the project and implemented on the spacecraft for several Titan and icy satellites flybys between September and October, 2005.