Recent studies have revealed that diverse urban geometry, as found in traditional urban areas, has positive consequences in ameliorating the urban microclimate in comparison to the more regular formal urban areas. The diversity in urban forms with variable street patterns, irregular plot sizes and building heights in traditional areas can help to lower the air-temperature (Ta) as well as mean radiant temperature (Tmrt) when compared to the formal areas. In contrast, formal or planned urban areas with uniform plot sizes and building heights, show a tendency to develop higher day-time temperatures. A field measurement in the tropical megacity Dhaka reveals, varying traditional urban forms are on average 1.70C cooler in comparison to more regular formal residential areas. The aim of this study is to understand the responsiveness of the microclimate simulation tool ENVI-met 4 in identifying the variation in urban geometry as reported in the previous study. The study aims to make specific comparisons between measured and monitored data by analysing a particular challenge in complex geometry, rather than develop a statistically significant large dataset. It attempts to demonstrate how ENVI-met could benefit from using the correct input as the boundary condition. While the modelling tool asserts to produce good results by using synoptic weather information as boundary conditions, this study suggests that it is important to use representative data from the actual site and that hourly input of climatic variables as boundary information can produce the best results. Results show that modelling is able to predict the relative variations in Tmrt conditions between sites, although highly overestimated. However, in terms of Ta, modelling was unable to produce any variations between different urban geometry characteristics. This indicates that, although ENVI-met can produce sufficiently good results in predicting Ta when hourly forcing is used, it is unable to distinguish between the precise details in urban geometry features that can cause significant variations in microclimatic conditions in real situations. Thus, more robust assessment of microclimatic variables is needed for using modelling techniques in order to evaluate the impact of diversity in urban geometry and subsequent microclimatic effects in urban canyons.