In a survey of 42 farm sites, comprising grassland and two types of horticultural farms (growing vegetables or flower bulbs), earthworm communities were sampled by hand-sorting and a number of soil physico-chemical characteristics recorded. For heavy metals the availability in the soil solution was estimated based on the measured absolute content. Abundance, biomass and species richness were significantly higher in grassland soils than in horticultural soils, and within the horticultural farms significantly higher in vegetable than in flower-bulb farms. No epigeic species were found in horticultural soils. The differences between the various farm types were probably related to the intensity of management practices, such as soil tillage, harvesting and crop protection measures, that result in less soil organic matter of lower quality. The data from the grassland farms were analyzed in conjunction with the physico-chemical soil characteristics, applying canonical correspondence analysis. Community parameters such as species richness and diversity apparently were not related to soil characteristics, but an appreciable part of the variation in abundance and biomass of several individual species could be explained by factors such as lutum content (particles <16 7m), pH, bio-availability of metal ions and geographical position of the sample site. It is concluded that repeated surveys of this type may become a valuable tool for monitoring soil quality.