Speeding is suited to investigations of risk-taking behaviour because there is a clear epidemiologic relationship between speeding and injurious crash risk (Kloeden, MacLean, Moore, & Ponte, 1997), crashing is an immediate negative outcome, and previous research has linked risk perceptions to speeding behaviour (Adams-Guppy & Guppy, 1995; Brown & Cotton, 2003). This investigation employed two exposure-free ways of measuring speeding risk perception; estimates of the danger inherent in exceeding speed limits in common speed zones, and generic beliefs about risk (Brown & Cotton, 2003). A 14 week prospective design was used to eliminate the possibility of reverse causality inherent in cross-sectional designs (Weinstein & Nicolich, 1993), and multivariate analyses were used to predict Time 2 (T2) risk perceptions statistically controlling Time 1 (T1) risk perceptions. The latter is a variation on the widely recognized strategy for the analysis of risk-perception data (Weinstein & Nicolich, 1993) and ensures consistency with Gerrard et al.’s (1996) methods. If risk-taking behaviour causes subsequently lowered risk perceptions, a negative association between T1 speeding behaviour and T2 risk perceptions should be observed. The extent to which changes in speeding behaviour between T1 and T2 predicted T2 risk perceptions was also examined. This was done for two reasons. Firstly, statistical control of T1 to T2 speeding changes improves the estimate of the T1 speeding-T2 risk perception association (Gerrard et al., 1996). For example, a driver might decrease speeding during this period, which could increase T2 risk perceptions. This would reduce the T1 speeding-T2 risk perception correlation. A similar increase in speeding would do the opposite. Secondly, there is some theoretical interest that speeding changes might also contribute to T2 risk perceptions (although reverse causality is also possible here). It was hypothesized that changes in speeding between T1 and T2 would also be inversely associated with lower risk beliefs at T2.