Context Urbanisation places increasing stress on ecosystem services; however existing methods and data for testing relationships between service delivery and urban landscapes remain imprecise and uncertain. Unknown impacts of scale are among several factors that complicate research. This study models ecosystem services in the urban area comprising the towns of Milton Keynes, Bedford and Luton which together represent a wide range of the urban forms present in the UK. Objectives The objectives of this study were to test (1) the sensitivity of ecosystem service model outputs to the spatial resolution of input data, and (2) whether any resultant scale dependency is constant across different ecosystem services and model approaches (e.g. stock- versus flow-based). Methods Carbon storage, sediment erosion, and pollination were modelled with the InVEST framework using input data representative of common coarse (25 m) and fine (5 m) spatial resolutions. Results Fine scale analysis generated higher estimates of total carbon storage (9.32 vs. 7.17 kg m−2) and much lower potential sediment erosion estimates (6.4 vs. 18.1 Mg km−2 year−1) than analyses conducted at coarser resolutions; however coarse-scale analysis estimated more abundant pollination service provision. Conclusions Scale sensitivities depend on the type of service being modelled; stock estimates (e.g. carbon storage) are most sensitive to aggregation across scales, dynamic flow models (e.g. sediment erosion) are most sensitive to spatial resolution, and ecological process models involving both stocks and dynamics (e.g. pollination) are sensitive to both. Care must be taken to select model data appropriate to the scale of inquiry.