Touch latency has been shown to deteriorate users' performances at levels as low as 25 ms, but this was tested only in short experimental sessions. Real life usage of touchscreens covers much longer periods. It provides training which could lead to reduce the impact of latency.We investigate users' ability to compensate for touch latency with training. Two groups of participants were trained on a tracking task during ten different days over two weeks with either high or low latency. The gap of performances between the two groups, observed at the beginning of the experiment, was reduced by 54 % after training. Users can thus compensate for latency, at least partially. These results nuance the negative effects of touch latency reported in previous work. They suggest that long-term studies could provide better insights on users' behaviors when dealing with touch latency.