Conservation tillage practices are included in the Finnish Agri-Environmental Program as phosphorus (P) loss control measures, but only few experiments have been performed to check their effectiveness in the local conditions. We studied surface and subsurface losses of P from a clayey underdrained field (Jokioinen/Kotkanoja; Vertic Cambisol/Typic Cryaquept), with 2% mean slope, during two separate experimental periods. Primary tillage treatments of the first experimental period of three years were moldboard ploughing (to 20-23 cm depth) vs. no autumn tillage (wintertime stubble). During the second experimental period of five years, the treatments were moldboard ploughing (20-23 cm) vs. shallow (to 5-8 cm) autumn tillage. The stubble treatment of the first experimental period produced higher dissolved reactive P (DRP) losses (104-259 g ha(-1) yr(-1)) than autumn ploughing (77-96 g ha(-1) yr(-1)), and equally high particulate P (PP) losses (mean 660, 235-1300 g ha(-1) yr(-1)). During the second experimental period, shallow autumn tillage produced 28% higher DRP losses (mean 120, 107-136 g ha(-1) yr(-1)) than ploughing (83-117 g ha(-1) yr(-1)) and 11% higher PP losses (mean 1090, 686-1336 g ha(-1) yr(-1)) than ploughing (783-1253 g ha(-1) yr(-1)). Surface runoff made up 28% and 16% of the total flow from the ploughed soil during the first and the second experimental period, respectively, as compared to 50% for the stubble and 44% for the shallow autumn tillage. Routing of flow between surface and subsurface pathways had a major influence on the P losses. In the relatively flat landscapes of the main agricultural areas of southern Finland, the potential for decreasing agricultural P losses by reduced tillage appears limited.