Most permafrost is located in the Arctic, and in total it contains around 600Gt of frozen organic carbon. This represents several times the remaining budget for anthropogenic emissions if we are to limit global warming below 2ºC, and this carbon could have a significant impact on global climate if it is released to the atmosphere following permafrost thaw. At present, the Arctic climate is changing much more rapidly than the rest of the globe, and yet observational data density in the region is low. The positive feedback between climate warming and permafrost carbon emissions depends on changing land-atmosphere energy and mass exchanges. There is thus a great need to understand links between the energy balance, which can vary rapidly over hourly to annual time scales, and permafrost, which changes slowly over long time periods. This understanding mandates long-term observational data sets. There is also a need to realistically incorporate permafrost into global modelling frameworks such as Earth System Models. Evaluating and parameterising these process-based models especially requires simultaneous measurements of interacting variables. Here we present an example of such a long-term data set, from the Bayelva Site at Ny-Ålesund, Svalbard, where meteorology, energy balance components and subsurface observations have been made for the last 20 years. Additional data include a high resolution digital elevation model and a panchromatic image. This paper presents the data set produced so far, explains instrumentation, calibration, processing and data quality control, as well as the sources for various resulting data sets. The resulting data set is unique in the Arctic and serves a baseline for future studies. Since the data provide observations of temporally variable parameters that mitigate energy fluxes between permafrost and atmosphere, such as snow depth and soil moisture content, they are suitable for use in integrating, calibrating and testing permafrost as a component in Earth System Models. The data set also includes a high resolution digital elevation model that can be used together with the snow physical information for snow pack modeling. The data show that mean annual, summer and winter soil temperature data from shallow to deeper depths have been warming over the period of record, indicating the degradation of permafrost at this site. This site is one of only a handful of permafrost sites where long-term automatic monitoring has taken place and data are accessible. There is a great need for continuous monitoring at more sites, to span the full range of permafrost conditions. Monitoring is often limited by scientific project funding, typically lasting only 3 or 4 years. This will continue to present a challenge unless there is investment by governments to operationalise these networks.