To provide a basis for the certification of wheat and maize flour reference materials containing deoxynivalenol, a series of three preliminary intercomparisons of methods was carried out between 1985 and 1988. The first intercomparison, involving 10 laboratories was designed to examine the determinative step and used a solution of deoxynivalenol in ethyl acetate and a spiked wheat extract solution. The second intercomparison involving 13 laboratories established minimum extraction times for naturally contaminated samples, demonstrated the absence of matrix effects in purified extracts and showed that there was no evidence of differences in the use of any of three different end determinations. The third intercomparison involving 15 laboratories used naturally contaminated cereals and acted as a preliminary to the final certification exercise. Over a period of four years there was a steady improvement in the range of results submitted by participants thought to be due to the application of improved analytical methods, increased expertise in the analysis and through experience gained through the intercomparisons.