The decision support system for the maintenance of paved roads, PMS, of the Swedish National Road Administration comprises a number of models for estimating the costs of road management authorities, road users and society at large. In 1998 VTI was commissioned by the Road Administration to carry out a comprehensive project with the objective of revising and augmenting these traffic effect models. The project is entitled "Traffic effects of the road surface" and has comprised a number of subprojects. The primary aim of this study has been to investigate, in a field study, the relationship between the roughness of the road surface (expressed in terms of the International Roughness Index IRI, with the units mm/m) and the perceived driving comfort of motorists, as well as their willingness to pay for improved driving comfort. 50 test subjects were asked to drive a car over nine 500 m long sections with different surface standards. The mean of the measured IRI on these road sections varies from 0.8 mm/m on the smoothest section to 10.5 mm/m on the roughest. In conjunction with the investigation, the test subjects were interviewed by a test leader. During the interview, they were asked which factors are significant for driving comfort and how they perceived driving comfort on the different sections. The state/condition of the asphalt surfacing was considered to have the greatest significance, followed by the car and the behaviour of other road users. Density of traffic was considered to have the least significance. The test subjects were also asked to say what significance different defects in the surfacing had on driving comfort. What is most important for a comfortable car journey is that there are no holes/depressions in the surfacing. This is followed by ruts, roughness that causes vibrations, and bumps and subsidence. The test subjects were asked to make a general assessment of driving comfort on a five-degree scale during their drive over the nine sections. A clear relationship was found between the assessment that the test subjects made and the measured IRI, i.e. the higher the IRI, the worse driving comfort was judged to be. The physical impact caused by the different surface standards of the test sections was also investigated by placing a dummy equipped with triaxial accelerometers in the passenger seat of the car that drove over the test sections. The willingness of the test subjects to pay for improved comfort associated with the condition of the road surface was also investigated by three methods: Stated Preference (Choice) and Contingent Valuation Method (driving comfort versus petrol price and trip time). A relatively high willingness to pay was obtained, but the methods gave very different results. It has therefore been impossible to posit a comfort cost model on the basis of these. This requires further studies and development of the methods.