Three different weed control treatments were applied every 3 weeks for 6 years to plots of Levin mottled silt loam previously under grass. The treatments were (i) spraying with a proprietary herbicide mixture containing equal parts of paraquat and diquat (0.3 kg total a.i./ha), (ii) spraying, plus hoeing to 1.5 em depth, and (iii) spraying, plus forking to 25 em depth. A mown grassed strip served as control. Soil density and aggregate stability were reduced by the treatments. Few large soil animals were present in any of the plots. The abundance and diversity of nematodes were reduced in the treatment plots; numbers declined significantly with soil depth only. in the grassed-strip and uncultivated plots, which had a vegetative cover of mosses and liverworts. A typical grassland microflora persisted in the treatment plots, although some bacterial groups, such as clostridia, were reduced in number. Soil from the treatment plots had lower organic C content and pH values, oxygen uptakes, carbon dioxide production, and dehydrogenase activity than soil from the grassed strip; dehydrogenase activity was significantly correlated with oxygen uptakes, and decreased significantly with soil depth. In general, the removal of the grass vegetative cover and the continued application of paraquat-diquat had most influence on the properties examined, and the additional weed control treatments had little effect.