The Urban Heat Island (UHI) phenomenon is associated to a cascade of effects on citizens’ health, energy consumptions and air quality. Accurate local scale UHI numerical predictions are particularly relevant for the achievement of a sustainable urban planning. In this paper, we evaluate the performance of two different modelling techniques to predict temperature distribution within the urban environment. The diagnostic integral-semi-Gaussian model ADMS-Temperature and Humidity Model and the prognostic Computational Fluid Dynamics (CFD) based model ENVI-met were used and results were validated against field measurements collected during summer 2012 in four different neighbourhoods of Lecce, an Italian medium size city. Results suggested that even if the entire building morphometry and green areas were carefully reproduced within the CFD-based model and only parameterized within the integral model, the level of complications appeared to be not essential in temperature predictions. As a consequence the integral model resulted to be more effective for a fast and accurate assessment of air temperature distribution within the city.