It is a fact that drivers’ poor speed choices play a significant role in crashes, leading to many fatalities and injuries worldwide (WHO, 2012). Young novice drivers are about twice as likely to be killed in a speed-related crash than older, more experienced drivers (NZTA, 2017). We also know that novice drivers’ hazard perception is often poor and predictive of crash likelihood (Horswill & McKenna, 2004), yet its relationship to their speed choices is virtually unknown. This thesis aimed to fill this critical gap. The first step (Experiment 1) was to examine the ecological validity of a new laboratory-based video speed choice task, which was similar to the task developed by Horswill and McKenna (1999). The speed choices of two gender-balanced groups comprising 24 ‘Novice’ drivers (mean age of 19.3 years) and 24 ‘Experienced’ drivers (mean age of 29.5 years) were recorded. Participants were shown video clips of various urban and rural road situations filmed from a driver’s perspective, and following each clip, asked to select the ‘appropriate’ speed they would feel most comfortable and safe travelling. An eye-tracker (SR-Research II) recorded their eye movements, which allowed for a detailed examination of drivers’ visual search behaviour. Compared to the Experienced drivers, the Novice drivers chose significantly faster overall speeds and adapted their speeds to a lesser degree under the differing road, weather, and lighting conditions. Novice drivers predominantly focused their visual attention immediately ahead of the simulated vehicle, with rapid glances at salient visual features, while the more experienced drivers focused their visual search more broadly and further ahead to include inspection of roadside cues. Road markings were also found to influence drivers’ speed choices, with the presence of clearly defined road-markings associated with higher speed choices. These laboratory-based results were consistent with what would be observed in real driving conditions based on data from naturalistic driving studies, real-world speed choice statistics, and crash data (Turner et al., 2014; Ministry of Transport, 2017). We concluded, therefore, that the speed choice task had considerable ecological validity. In Experiment 2, we replicated the speed choice task conditions of the first experiment but added a separate video-based hazard perception task (Isler, Starkey, & Williamson, 2009), and tested 138 participants, divided into five gender-balanced groups based on age, experience, and licence type (mean age: ‘Learner’= 16.5 years, ‘Restricted’= 18.8 years, and ‘Full (<25)’ licence= 23.2 years; ‘Full (25<50)’= 34.9 years, ‘Full (>50)’= 57.5 years). Our prediction, based on the reviewed literature (e.g., McKenna, Horswill, and Alexander, 2006), was that more advanced hazard perception skills would facilitate increased awareness of risk, prompting the selection of slower appropriate speeds. The results indicated that both the number of perceived hazards and hazard perception times significantly improved with experience as anticipated. Drivers’ chosen speeds, however, increased with age, with ‘Experienced’ drivers choosing faster speeds than novice drivers (‘Learner and Restricted’) and, to a lesser degree, ‘Full (>50)’, older drivers. This indicated that higher levels of hazard perception skills were often related to choices of faster speeds in the speed choice task, and this finding was unexpected. We concluded that it might be possible that experienced drivers only select slower speeds at the time when they become aware of immediate hazards. This hypothesis required further clarification, forming the basis of the rationale for conducting Experiment 3. In Experiment 3, the same hazard perception task as in the second experiment was merged with the speed choice task to form an experiment measuring speed choices under the immediate influence of hazards. Two groups of participants, 52 ’Novice’ drivers (mean age of 19.9 years) and 37 ‘Experienced’ drivers (mean age of 37.4 years), were asked to select the speed they considered most appropriate immediately following each hazard perception trial. Visual search patterns were recorded using an eye-tracker. This time, longer hazard perception times were associated with choices of faster speeds. Overall, novice drivers showed less efficient visual search strategies when perceiving hazards, requiring about double the number of fixations to identify each hazard, but they also chose faster speeds than their more experienced counterparts. We concluded that if there is a causal relationship between improved hazard perception skills and speed choices, we might reduce drivers chosen speeds in Experiment 4 by improving hazard perception, particularly in novice drivers. Experiment 4 consisted of two studies. In the first study, forty participants were randomly assigned either to a control group or a training group, each composed of twenty drivers ‘Novice’ (mean age of 16.6 years) and ‘Experienced’ drivers (mean age of 31.1 years). Both Novice and Experienced drivers showed significant improvements in hazard perception following road commentary, accompanied by a change in visual focus, compared to the control group. In the second study, we tested the hypothesis that improved hazard perception will lead to slower speed choices. Twenty-two participants, ten ‘Novice’ drivers (mean age of 21.3 years) and twelve ‘Experienced’ drivers (mean age of 29.1 years), were assigned randomly to either a control or test group. We found that immediately following the road commentary, the test group showed significant improvement in hazard perception skills and chose slower speeds compared to the control group. In summary, this thesis revealed several significant new findings and insights, leading to a much better understanding of the underlying factors influencing speed choices of novice and experienced drivers. More efficient hazard perception was clearly related to choices of slower speed when hazards were presented within the speed choice trials, possibly mediated by visual search behaviour. There was strong evidence of a causal relationship when road commentary improved hazard perception and caused drivers to select slower speeds, directly influencing speed choices. Future research could investigate the potential for hazard perception to reduce speed choice in real-world traffic situations, especially for young novice drivers. The knowledge that improving hazard perception can influence safer speed choices is of great value for future road safety initiatives. Such research may be instrumental in the quest to decrease the number of speed-related crashes both in New Zealand and around the world.