Abstract In 2016, the Royal Dutch Meteorological Office (KNMI) homogenized the daily temperature records for the Netherlands from 1901 to 1950 to allow a realistic comparison of the temperatures from 1901 to the present. The homogenizations for the main station De Bilt were carried out using a Percentile Matching Method (PMM) with one reference station and a 56-month reference period. In this study, it is shown that the corrections in the number of tropical days (maximum temperature ≥ 30 °C) depend strongly on the choice of the reference station and the length of the reference period. A total of 116 different variants of the homogenization of De Bilt were carried out, using all combinations of five reference stations, five reference periods, two ways to calculate percentiles, and two ways to smooth the data. The parameters used for the KNMI’s current homogenization of De Bilt result in a very sharp decrease of tropical days, which is not replicated by the majority of the 116 variants. Moreover, after homogenization, De Bilt appears to be an outlier compared to the other meteorological stations. Therefore, the current homogenized estimates of tropical days for De Bilt should be treated with considerable caution.