One way to estimate the temporal precision of vision is with judgments of synchrony or temporal order of visual events. We show that irrelevant motion disrupts the high temporal precision that can be found in such tasks when the two events occur close together, suggesting that the high precision is based on detecting illusory motion rather than on detecting time differences. We also show that temporal precision is not necessarily better when one can accurately anticipate the moments of the events. Finally, we illustrate that a limited resolution of determining the duration of an event imposes a fundamental problem in determining when the event happened. Our experimental estimates of how well people can explicitly judge when something happened are far too poor to account for human performance in various tasks that require temporal precision, such as interception, judging motion or aligning moving targets spatially. © 2010 Elsevier Ltd.