Fifty percent of drives have been found to frequently violate speed limits (OECD/ECMT, 2006), and speed limit compliance appears to depend on the perceived credibility of a road’s speed limit (Goldenbeld & van Schagen, 2007). Credibility of speed limits, in turn, appears to be determined by the match between drivers’ speed preferences and the design of the road (Goldenbeld & van Schagen, 2007). Yet, a challenge has been that not all drivers’ prefer the same speeds, and individual differences with regards to speed preference lead to speed variability and speed conflicts in traffic (Elvik, 2010). The aim of this thesis was to explore whether the speed drivers like to drive when motivated by different driving goals (speed preference) correspond to the speeds that they actually drive on those same roads (speed choice). Additionally, this thesis sought to explore the relationship between speed preference and risk perception. Data was collected in two ways, from a speed gun and from a questionnaire. The speed gun collected on-road measures of driving speeds on seven different roads, while the questionnaire collected measures of drivers’ self-reported speed, speed preference and risk perception. For the speed preference measures, participants were asked what speed they would choose on a given road when: 1) motivated by safety, 2) considering fuel savings, or 3) motivated by fun, and additionally 4) what speed they usually drove on the road. In total 200 drivers were interviewed at five different parking lots, and they referred to the seven roads that were sampled with the speed gun. The results indicated that speed preference helped to explain actual driving speeds. More specifically, drivers’ different driving goals and their large individual differences with regards to speed preference corresponded to different speed choices. No relationship was found, however, between drivers’ speed preferences and their risk perceptions. The results are discussed with regards to implications for the problem of speed variability in traffic.