Abstract During 1954–1957 particles of irradiated uranium oxide were emitted from the Windscale Piles. The particles were large compared with most ambient aerosols, 85% by weight being between 10 and 100 μm diameter, and they mostly fell within a few kilometers of the works. From measurements of activity in soil in 1958 onwards, it is estimated that about 12 kg of uranium, 20 Ci of 137Cs, 18 Ci of 90Sr and 0.3 Ci of 239+240Pu were emitted. To estimate the concentrations of 90Sr and 137Cs in milk during the emissions, comparison is made with the fallout of fission products from bomb tests, which also began in 1954, but continued longer. It is calculated that the average level of 90Sr in milk at Seascale, 1954–1957, was 36 pCi per g Ca, with higher concentrations, 100–200 pCi (g Ca) −1, in milk from farms immediately surrounding the Windscale plant. For comparison, in 1963/1965 the average level of 90Sr in milk in the UK was 20–30 pCi (g Ca) −1, with 90–170 pCi (g Ca) −1 in milk from certain hill farms. Since the oxide particles were larger than those derived from bomb tests, and the conditions of deposition different, an extended review is given (Appendix 1) of the factors determining the entry of radioactive or stable contaminants from fallout into food chains. It is concluded that the oxide particles would have been less efficient in this respect than bomb fallout, mainly because their relatively large size and high density would have given impact velocities sufficient to cause them to bounce off leaves, leaving little activity on the edible herbage. For this reason, the calculations are on the safe side.