The key parameters describing the production and disposal of dirty water on 20 dairy farms in England and Wales were investigated. Farms were selected where existing pumped dirty irrigation systems allowed for appropriate sampling and monitoring. Each farm was visited to identify sources of dirty water, quantities and types of dairy chemicals used, details of storage and settlement tanks and current field irrigation techniques. Dirty water arose from five main sources: bulk milk tank rooms, milking parlour pits, milking parlour cow standing areas, outdoor fouled yards and silage cramps. On 14 farms, slurry or drainage from farmyard manure stores also entered the dirty water system. The volumes of dirty water produced varied greatly with no simple correlation with the number of cows, suggesting that factors such as rainfall and outdoor yard areas had a major influence. The average annual application rates to land varied from 29 to 817 m(3) ha(-1), with the heaviest applications on undrained light soils. These findings, together with biochemical characteristics, indicated that effective biological treatment for dirty water was feasible but the data describing dairy chemical usage showed that significant concentrations of potential bactericidal materials were likely to arise. This could seriously impair such treatment. The effects of volumetric buffering were estimated, showing how this could improve the potential reliability of treatment. It was concluded that pilot-scale development would be an essential first step in designing a practical treatment system.