Walkable access is recognised as one of the most important factors for deciding to walk instead of using other modes of transport. Distance has been less accurately taken into consideration in previous walking accessibility measures, however, as they are often based on an isotropic approach or on a fixed distance threshold. The objective of this paper is to present a method of modelling continuous walking accessibility to different amenities in a city, with an integrated network-based and distance-decay approach, applied to a case study of the city of Ljubljana, Slovenia. The approach is based on a web survey to obtain data on acceptable walking distances to different types of amenities. Several distance decay functions were analysed for each type of amenity from the cumulative frequency of responses. The best fitting functions were used to model the walking accessibility surfaces for individual amenities in the network, representing five domains (retail, services, recreation, education and transportation) and an overall walking accessibility index. Despite certain limitations and a further need to assess the validity of the methods, our distance-decay network-based approach is more accurate than the isotropic or even network-based modelling of walking distances in continuous or threshold approaches, as it enables the researcher to take into account the differences in propensities to walk to different amenities. The results can be used by city authorities and planners for implementing actions to improve walking accessibility in the most problematic areas.