Migration of Cs-137 and Sr-90 in undisturbed soil was studied in large lysimeters three and four years after contamination, as part of a larger European project studying radionuclide soil-plant interactions. The lysimeters were installed in greenhouses with climate control and contaminated with radionuclides in an aerosol mixture, simulating fallout from a nuclear accident. The soil types studied were loam, silt loam, sandy loam and loamy sand. The soils were sampled to 30-40 cm depth in 1997 and 1998. The total deposition of Cs-137 ranged from 24 to 45 MBq/m(2), and of Sr-90 from 23 to 52 MBq/m(2). It was shown that migration of Cs-137 was fastest in sandy loam, and of 90Sr fastest in sandy loam and loam. The slowest migration of both nuclides was found in loamy sand. Retention within the upper 5 cm was 60% for both C-137 and Sr-90 in sandy loam, while in loamy sand it was 97 and 96%, respectively. In 1998, migration rates, calculated as radionuclide weighted median depth (migration centre) divided by time since deposition were 1.1 cm/year for both Cs-137 and Sr-90 in sandy loam, 0.8 and 1.0 cm/year, respectively, in loam, 0.6 and 0.8 cm/year in silt loam, and 0.4 and 0.6 cm/year for Cs-137 and Sr-90, respectively, in loamy sand. A distinction is made between short-term migration, caused by events soon after deposition and less affected by soil type, and long-term migration, more affected by e.g. soil texture. Three to four years after deposition, effects of short-term migration is still dominant in the studied soils. (C) 2000 Elsevier Science Ltd. All rights reserved.