As buildings are main contributor to greenhouse gas emissions, it is important to assess the performance of existing buildings and assist the design of new sustainable buildings through building energy simulation. It is well known that by using local climate measurements for building energy simulation would provide more accurate result than by using other typical weather data, i.e. typical meteorological year (TMY). However, as different built forms/architectural layouts would also have impacts on neighbourhood-scale microclimate, it is worthy to quantify the difference it would make. In this study, we performed a year-long measurement with four weather stations surrounding a campus building in 2009 and 2010. Each station was placed in a typical type of built form, including a street canyon, a courtyard, a semi-closed courtyard and a relatively larger open area. Besides, two typical weather data files, typical meteorological year (TMY) and actual meteorological year (AMY) were taken as reference. Annual heating demand and natural ventilation cooling potential were calculated based on all 6 weather files. Our simulation results show that the variation in annual heating demand of different built forms could be between 1.1 and 7.3%, where the large open area has the highest heating demand and it of the courtyard is the lowest. The difference between on-site measurement and TMY in annual heating load is as high as 10.8%. While in summer, night ventilation cooling potential of the courtyard and the semi closed form are the highest, and it of the street canyon is the lowest. Using TMY could underestimate the night ventilation cooling potential by 26-31% and using AMY could overestimate it by 9-14% in total. Overall speaking, the courtyard form shows good performance in reducing heating demand and enhancing night ventilation cooling, while the street canyon shows relatively poor performance in both aspects. These findings highlight the importance to understand the impact of neighbourhood-scale microclimate on building energy performance. (C) 2019 Published by Elsevier Ltd.