The negative of the vertical component of the atmospheric direct current (DC) electric field is referred to as the atmospheric electric potential gradient (PG). The PG depends on the actual ionospheric potential, local electric fields, and the electrical conductivity of the air at the place of the measurement. These factors are more or less directly connected to meteorological conditions. The overall state of the global network of large-scale electrical currents in the Earth-ionosphere system can be inferred from the PG when the weather is locally calm. This meteorological condition is traditionally referred to as “fair weather” and is characterized by allowed ranges of specified meteorological parameters. This is why information on the actual weather condition is supposed to be an inherent supplement of PG datasets.This dataset contains PG data recorded in the Széchenyi István Geophysical Observatory (NCK, 47.632°N, 16.718°E), Hungary from 1962, when the regular measurements were started, up to 2009. Throughout this time period, data were collected using practically the same instrument at the same location. The PG was measured by a locally developed radioactive apparatus which equalizes the atmospheric potential over the lowest 1 m thick air layer so that the potential difference between the sensing and grounded electrodes at ground level is the PG itself. Zero signal offset was determined daily and the instrument was calibrated in the ±250 V/m range weekly whenever it was possible. The instrument has a measuring range of −300 V/m to 300 V/m. The data were recorded photographically by a sensitive galvanometer. Hourly averages were then obtained from the photographical records via manual evaluation with an uncertainty of ±10 V/m. Hourly averaged PG data was included in this dataset when valid records from more than 30 minutes from a given hour were available. Records in the original dataset marked as unreliable or saturated have been omitted. Detailed characteristics of the instruments and the applied calibration technique as well as links to original data publications can be found in Bór et al., 2020 and the references therein.This dataset also contains hourly PG averages which have been corrected for the time-dependent bias caused by the electrostatic shielding effect of trees that were growing up not far from the measuring instrument over the decades. Note that this shielding effect largely dominates the long-term trends in the uncorrected data, so the original PG data must be interpreted with care. The uncertainty of the conversion is also provided. This uncertainty arises from unexact information on the age and growth rate of different trees near the measuring instrument. Detailed explanation of the correction can be found in Buzás et al., 2021.On-site measurements of temperature, total rainfall, relative humidity, resultant wind direction and speed, and global solar radiation are available after 2000. These parameters were measured by a Campbell meteorological station. The measured data was compiled using the factory calibration of the sensors throughout the included time interval. The original time resolution of the data is 10 minutes. The data was converted to hourly time resolution to comply with the PG records.In order to provide a possibility for examining the relation of PG to meteorology on the full time span of the PG records, meteorological parameters obtained in hourly time resolution from the ERA5 reanalysis framework are attached. The two horizontal components of wind speed at 10 m (to calculate the resultant wind speed and direction; see reference ERA5WindCalculation), temperature at 2 m, dew point temperature at 2 m (to calculate relative humidity according to Tetens (1930)), total precipitation, surface pressure, downwards surface solar radiation, snow cover, and snow depth were extracted from the ERA5-Land hourly data 1950 to present dataset (Muñoz Sabater, 2019). Total cloud cover, low cloud cover, and cloud base height between 1962 and 1978 were compiled from the ERA5 hourly data on single levels 1950-1978 (preliminary version) dataset (Bell et al., 2020). The latter parameter set in the 1979–2009 time period was compiled from the ERA5 hourly data on single levels 1979 to present dataset (Herschbach et al., 2018). Note that, due to the finite spatial resolution of the ERA5 framework, these values correspond to the whole area 47.6-47.7 °N latitude and 16.7-16.8 °E longitude (11 km x 11 km) in the case of the ERA5-Land hourly data dataset and 47.5-47.75 °N latitude and 16.5-16.75 °E longitude (28 km x 28 km) in the cases of ERA5 hourly data on single levels datasets. The observatory is located 4 km and 13 km from the center of the corresponding area in the case of higher and lower spatial resolution ERA5 datasets, respectively.