Nitrogen (N) fixation by white clover (Trifolium repens L.) was measured in the ley phases of three experimental organic ley-arable rotations at two sites, between 1997 and 2000, using the 15N dilution technique. Soil nitrate-N, ammonium-N, and total soluble N were measured at one of the sites between 1999 and 2001. Variation of soil N was assessed in different ages of ley, year, month, rotation, site and microsite type (grass or clover). Data for soil pH, organic matter, P, K, and Mg, was also compared, as was weather data for the period. Variation of N fixation, grass and clover dry matter, proportion of nitrogen derived from the atmosphere, grass and clover N concentrations, grass and clover non symbiotic N-yield, and N transfer from clover to grass were compared in different ages of ley, years, months, rotations and sites. Flowering rates and presence of clover tap-roots were also examined. Above ground N fixation varied between 26 and 75 kg ha'1 over the course of the growing season. Estimated total N fixation ranged from 70-214 kg ha'1. N fixation was closely related to yield of clover. There were clear relationships between soil nitrate and the behaviour of the grass and clover components of the leys. Nitrate-N was slightly higher under clover patches than under grass patches in mid-late summer. White clover and perennial ryegrass were grown in circular chambers, divided radially, and their invasion into microsites with different vegetation, different management and different N treatments was observed. Ryegrass was not capable of rapidly colonising microsites by vegetative spread alone. Clover was more likely to successfully invade microsites containing grass with no added N than microsites with added N, or bare soil microsites.