Aims. Over the past few years a major effort has been put into the exploration of potential sites for the deployment of submillimetre astronomical facilities. Amongst the most important sites are Dome C and Dome A on the Antarctic Plateau, and the Chajnantor area in Chile. In this context, we report on measurements of the sky opacity at 200 μm over a period of three years at the French-Italian station, Concordia, at Dome C, Antarctica. We also present some solutions to the challenges of operating in the harsh polar environment. Methods. The 200-μm atmospheric opacity was measured with a tipper. The forward atmospheric model MOLIERE (Microwave Observation LIne Estimation and REtrieval) was used to calculate the atmospheric transmission and to evaluate the precipitable water vapour content (PWV) from the observed sky opacity. These results have been compared with satellite measurements from the Infrared Atmospheric Sounding Interferometer (IASI) on Metop-A, with balloon humidity sondes and with results obtained by a ground-based microwave radiometer (HAMSTRAD). In addition, a series of experiments has been designed to study frost formation on surfaces, and the temporal and spatial evolution of thermal gradients in the low atmosphere. Results. Dome C offers exceptional conditions in terms of absolute atmospheric transmission and stability for submillimetre astronomy. Over the austral winter the PWV exhibits long periods during which it is stable and at a very low level (0.1 to 0.3 mm). Higher values (0.2 to 0.8 mm) of PWV are observed during the short summer period. Based on observations over three years, a transmission of around 50% at 350 μm is achieved for 75% of the time. The 200-μm window opens with a typical transmission of 10% to 15% for 25% of the time. Conclusions. Dome C is one of the best accessible sites on Earth for submillimetre astronomy. Observations at 350 or 450 μm are possible all year round, and the 200-μm window opens long enough and with a sufficient transparency to be useful. Although the polar environment severely constrains hardware design, a permanent observatory with appropriate technical capabilities is feasible. Because of the very good astronomical conditions, high angular resolution and time series (multi-year) observations at Dome C with a medium size single dish telescope would enable unique studies to be conducted, some of which are not otherwise feasible even from space.< Réduire