A replicated, long-term field drainage experiment on a clay soil was started in 1978 to investigate the effects of mole drainage on land that had been direct-drilled or ploughed. We here describe the background and the need for this work, details of the experiment and summarize the results for the first 2 years. The main factors causing point-to-point variability in the soil at the experimental site and their relative contributions were assessed. Variations in inherent soil properties were small, except for depth to calcareous clay. Site hydrology was also generally uniform, especially soil water content, and was little affected by the presence of subsoil drainage schemes dating from the 19th century. There were some variations in micro-relief. The drained and undrained plots were hydrologically separated. Equipment was installed to measure separately surface runoff, lateral flow at the bottom of the cultivated layer (the interflow) and deep drainage from the mole and pipe system (in the drained plots only). The design and performance of that equipment is described. In both years much of the undrained soil became saturated in winter, but a residual cultivation pan limited the effect of the mole drains in 1978–9. In 1979–80, after disrupting the pan, the water table in the moledrained plots was about 25 cm deeper than in the undrained plots, root growth was greater and yield of winter wheat 11% heavier. About 90% of water and nutrients draining from the drained plots were carried in the mole and pipe system. The results on water balance, nutrient losses, crop growth and yield are fully reported in companion papers (Harris et al. 1984; Ellis et al. 1984).