Temperatures of ten thermometer screens have been studied for particular weather conditions during a 6-year field experiment in De Bilt (the Netherlands). The comparison comprised two versions of an aspirated Young screen, 4 naturally ventilated round-shaped multi-plate screens (KNMI, Vaisala, Young, Socrima), a slightly aspirated version of the KNMI screen, a synthetic Stevenson screen (both aspirated and naturally ventilated) and a naturally ventilated wooden Stevenson screen. All screens were equipped with fast-responding sensors. A simple method is presented to obtain inter-sensor accuracies of about 0.03 degr.C under field conditions. The response time of the screens is studied by making a daily comparison of the timestamps of the minimum and maximum temperatures. The analysis shows that the response of the naturally ventilated Stevenson screens is about 7–8 minutes slower than for the other screens. The screens have been compared for conditions of rainfall, wind, clear sunny days, days with snow cover and days with fog. It is demonstrated how these weather conditions affect the temperature measurements of the screens. The results show that the screens can roughly be classified into three distinct groups: (1) the round-shaped multi-plate screens, (2) the naturally ventilated Stevenson screens, and (3) the strongly aspirated screens. Due to the complexity of reducing climate and siting dependent environmental impacts on temperature measurements, it is not possible to design one particular screen as a world-wide reference. For each climate a special screen has to be developed as the best balance between the application of impacts reduction techniques and sensing the real air temperature.