Abstract: The K-means clustering method was used to determine quality-controlled hourly surface air temperature data of 123 automatic weather stations in the Beijing area. The sites of these stations were classified into the following four zones, that is urban stations, suburban stations, west and north mountainous area stations, and southwest and northeast mountainous area stations. Monthly and diurnal temperature variations and diurnal temperature range (DTR) of the four zones were analyzed and hourly and monthly urban heat islands were investigated. K-means clustering proved advantageous in classifying temperature zones, which were correlated with the terrain and underlying surfaces of the stations. The DTR of the urban zone was revealed to be smaller than that in suburban and mountainous areas. The west and north mountainous areas exhibited the highest DTR values. All zones exhibited high DTR values in February, May, and October. The urban heat island effect in the Beijing area was stronger in winter and at night. In the daytime, the urban heat island effect appeared to be stronger in summer than that in other seasons. From March to August, the urban heat island effect was the weakest.