Calli of resistant, intermediary and susceptible wheat (Triticum aestivum L.) varieties were selected using culture filtrates of Fusarium graminearum and F.culmorum and the regenerants were evaluated for resistance up to R(3). Czapek-Dox broth medium was inoculated with mycelia of Fusarium isolates and incubated for 2-6 weeks. Filtrates were added to MS callus growing medium, then 5 weeks-old calli were transferred onto this medium (MST) for 4-5 weeks. MST containing 30% filtrate was found to be suitable for selection. Resistant calli were transferred again to fresh MST for further two selection cycles. The surviving calli produced less fertile regenerated lines (R(0)) than the non-selected ones. Among 18 R(1) lines tested for Fusarium-resistance in the seedling stage by artificial inoculation in the greenhouse, two (11.1%) were significantly more resistant, one (5.6%) was more susceptible than the original cultivar and the rest (83.3%) behaved similarly to the donor plants. Among unselected R(3) lines of three varieties, practically the same number of resistant plants were found as among the related selected ones. When the R(3) selfed generations obtained through double-layer and culture filtrate selection techniques were tested for Fusarium-resistance, 35.7% of the lines were found to be more resistant than the original cultivars, none was more susceptible and 64.3% had a reaction similar to that of the source materials. Thus, inheritance of the disease reaction was not stable in all cases. Success of in vitro selection for Fusarium-resistance depended also on the genotype, and toxin analysis showed that although being effective, the selective media contained deoxynivalenol only exceptionally. In selecting wheat for Fusarium-resistance in vitro, the culture filtrate technique proved better than the double-layer procedure.