In the multidisciplinary ALentejo Observation and Prediction systems project (ALT20-03-0145-FEDER-000004), which aims to strengthen research and innovation in the Alentejo region (southern Portugal), one of the main objectives was to study and model the meteorological conditions in the Alqueva reservoir, in particular their spatial variations within a few hundred meters. The shared hourly dataset, using Coordinated Universal Time (UTC), covers the period from 2018 to 2023 and includes measurements from two meteorological stations located in the Alqueva reservoir, the largest artificial lake in Europe. One station, Montante, is located on a floating platform with a water depth of approximately 70 meters (38.2235 N, 7.4595 W), to the west of the second station, CidAlmeida (38.21539 N, 7.45454 W), which is about 1 km away on land, very close to the water. According to the World Meteorological Organisation (WMO) standards, the data were sampled every second, and hourly data were calculated in post-processing. The dataset includes hourly accumulated precipitation (mm) and hourly average measurements of surface water temperature (at a depth of 0.25 m), soil temperature (at a depth of 0.15 m) and various meteorological parameters: wind speed (m/s) and direction (degrees), relative humidity (%), upward/downward solar radiation (W/m2 ) and air temperature (°C ). All parameters are measured at both stations, except the hourly average water temperature at a depth of 0.25 m, which is only available at the Montante station, and the hourly accumulated precipitation, hourly average soil temperature and wind direction, which are only available at the CidAlmeida station. Hourly data were not subjected to rejection criteria based on the percentage of errors; instead, a column with this percentage is provided, allowing potential data users to apply their own rejection criteria. Daily extremes (daily maximums and minimums for air temperature, relative humidity, and daily maximum gust) are only provided for days where the percentage of errors does not exceed 25% of the 1440 minutes of each day. The daily error percentage for each of the measured parameters at the two stations for the period 2018-2023 is available. Finally, the repository also includes two codes (one for each weather station) written in Visual Basic, which enable data transmission and real-time statistical processing. Fundings: Gonçalo Rodrigues was supported by the Portuguese Foundation for Science and Technology, I.P (Grant 2020.05752.BD). The work is co-funded by national funds through FCT – Fundação para a Ciência e Tecnologia, I.P., in the framework of the ICT project (references UIDB/04683/2020 and UIDP/04683/2020) and by the ALOP project (ALT20-03-0145-FEDER-000004).