Abstract Climate data quality concerns prompted this study. The objective was to ascertain the impact of air temperature inhomogeneities on derived weather variables, such as growing-degree days, frost dates and heat stress, and to develop adjustment methodologies. Daily weather data spanning 1893–1989 from 34 Iowa stations were used for computation. Direct comparisons for each weather station and derived variable were made between all possible weather station combinations and a geographical mean data set. The latter was computed by using all available weather stations. For all derived weather variables studied, most weather stations revealed some type of notable inhomogeneity over the long period. Homogeneous weather stations were indeed rare. Individual time series exhibiting different levels of stability were common as were those exhibiting sharp discontinuities. Many, but not all, discontinuities could be traced to station histories. The use of the geographical mean data sets provided a rapid method to screen candidate weather stations, and could be used to correct discontinuities if they were not too numerous. More precise adjustments could, however, be made by using one or two nearby weather stations that were identified as being extremely stable. This process was more tedious because the stable periods were usually short. Climate resources are a valuable asset for mankind. However, data quality needs to be scrutinized before subsequent climate analyses.