Different variables concur to the outdoor comfort assessment and, among them, the Mean Radiant Temperature (Tmr) represents the most challenging one to experimentally evaluate. According to the scientific literature, the most accurate method for in-field Tmr calculation is integral radiation measurements with three net radiometers. Nevertheless, net radiometers are expensive sensors and their implementation on large scale may be hindered by their price. To provide a cost-effective alternative some researchers proposed the use of globe thermometers. Globe thermometers are affordable sensors but typically exploited indoors, and their accuracy in urban settings is still under investigation. The scope of this work is to provide information regarding the extent to which globe thermometers can substitute net radiometers in the evaluation of the Tmr in outdoor urban settings. To this purpose, an experimental comparison has been performed between the Tmr calculated using net radiometers, assumed as a reference, and using two different globe thermometers: a standard 150 mm black globe and a grey 50 mm globe. The results revealed that the black globe tends to overestimate the Tmr while the grey globe mainly underestimates it. Moreover, the analyses evidenced a wide fluctuation in the profile of the Tmr calculated using globe thermometers, due to the rapid variation of meteorological parameters and the globes' long response time. To improve the fit between globe thermometers and net radiometers' response, regressive functions have been proposed, resulting in a significant enhancement of the adherence between Tmr calculated using the standard black globe's measurements and the reference Tmr. •Tmr is key parameter for outdoor thermal comfort assessment in urban settings.•Globe thermometers cannot directly substitute net radiometers to assess the Tmr.•A regression equation is used to assess the Tmr from the standard globe thermometer.•In 90.4% of cases, the equation results in a negligible error for Tmr.•This result decreases greatly the cost and number of required sensors.