Root biomass production, root length production and root turnover of Erica tetralix and Molinia caerulea were estimated by sequential core sampling and by observations in permanent minirhizotrons in the field. Root biomass production, estimated by core sampling, was 370 (Erica) and 1080 (Molinia) g m-2 yr-1. This was for both species equal to aboveground production. Assuming steady-state conditions for the root system, root biomass turnover rates (yr-1), estimated by core sampling, were 1.72 (Erica) and 1.27 (Molinia). Root length production of both species, estimated by minirhizotron observations, varied significantly with observation depth. Root length turnover rate (yr-1) of both species did not vary significantly with observation depth and averaged 0.92 in Erica and 2.28 in Molinia. Reasons are given for the discrepancy between the results of the two types of turnover measurements. The data suggest that the replacement of Erica by Molinia in a wet heathland, which occurs when nutrient availability increases, leads to an increased flow of carbon and nutrients into the soil-system. Therefore, there may be a positive feedback between dominance of Molinia and nutrient availability.