In recent years a range of new methods have been proposed with which to describe and evaluate driver behaviour. One such method is that of fuzzy logic, where variables used in the driver decision-making process may be described linguistically, allowing a quantifiable degree of uncertainty to be introduced. This paper explores the use of such a formalism to describe the driver perception of 'closing speed' between two vehicles engaged in 'car-following' on a motorway, and by using data from an instrumented vehicle experiment, it tests a number of models using relative speed, visual angle and the time to collision. Several of these models fit the data quite well, and there is both a small positive perception bias present and a number of reversals in sign judgement. Additionally, a brief examination is made of potential variations on the methodology that may both make data collection easier and/or allow a 'more fuzzy' representation to be made.