Until recently, the design of road infrastructure involved mainly concerns related to the base speed value and to requirements ensuring a high level of service. Currently, it is widely accepted that only an integrated approach is able to take into account the interests and needs of all the involved stakeholders, namely the need of traffic fluidity by motor vehicles, and quality of life and safety associated to other users and local residents. This vision has been leading to different speed management approaches, particularly in relation to the setting of the appropriate and coherent speed limits along extensive routes. In this context, this study is part of a broader research program aimed to develop a methodology of speed management with a widespread use applied to single carriageway roads in interurban areas, crossing different road surrounding environments with varying rural, urban or mixed characteristics. More specifically, this research line is focused on the development of a decision-support methodology for the definition of the appropriate speed limit, based on the drivers’ willingness to naturally accept successive changes in the speed limits in successive stretches of the roads. This decision-making process necessarily involves a broad set of variables and factors that might be divided in two fundamental components: (1) road geometry, and (2) interaction between the road and its surrounding environment. The effect related to the geometric characteristics has already been extensively studied, thus, this paper is focused on speed limit modelling having exclusively into account the latter component, leaving geometryrelated considerations to the final decision-making process. A MultiNomial Logit (MNL) discrete choice model was developed. The modelling work was carried out using a case study involving 55.4 km and 11 km more, for validation purposes – of rural roads crossing different environments. The model was estimated based on the judgments of fourtraffic safety experts applied to each direction of each road stretch (200 m long). Before the modelling development, a factor analysis involving the whole set of variables was carried out, in order to detect not directly observable common features on its structure that may influence the final outcome of model estimation. This analysis allowed to corroborate the overall variable selection process. The developed model adjusts quite well to the data with McFadden pseudo-R2 of 0.447. Without loss of explaining power it was possible to avoid the presence of subjectivity related to a qualitative evaluation of some explanatory variables, greatly enhancing the model’s robustness and transferability to other locations and countries. The new modelling structure also facilitates the interpretation of the validation results and enables the models to be developed to represent different levels of safety related sensitivity existent amongst the expert community, by using a ‘conservative expert’ evaluation. The results validation has showed the models to be robust and useful as decision support tools applied to speed limits’ evaluation processes.