Previous work on ocular-following emphasises the accuracy of tracking eye movements. However, a more complete understanding of oculomotor control should account for variable error as well. We identify two forms of precision: ‘shake’, occurring over shorter timescales; ‘drift’, occurring over longer timescales. We show how these can be computed across a series of eye movements (e.g. a sequence of slow-phases or collection of pursuit trials) and then measure accuracy and precision for younger and older observers executing different types of eye movement. Overall, we found older observers were less accurate over a range of stimulus speeds and less precise at faster eye speeds. Accuracy declined more steeply for reflexive eye movements and shake was independent of speed. In all other instances, the two measures of precision expanded non-linearly with mean eye speed. We also found that shake during fixation was similar to shake for reflexive eye movement. The results suggest that deliberate and reflexive eye movement do not share a common non-linearity or a common noise source. The relationship of our data to previous studies is discussed, as are the consequences of imprecise eye movement for models of oculomotor control and perception during eye movement.