Experiments were carried out to study the differential responses of different wheat cultivars to boron toxicity in field, greenhouse and growth chamber conditions. In field trials carried out at two locations, both of which are known to contain toxic amounts of water-extractable B, significant correlations were obtained between toxicity symptoms and grain yields. The only durum cultivar included in this group of experiments (Kunduru 1149) was the most sensitive of the 21 cultivars trialed. The most tolerant cultivars were of local origin. Genotype-environment interaction was considerably large. Twenty-nine bread wheat and three durum wheat cultivars were compared in a greenhouse experiment with and without the application of 40 mg L(-1) B. Again, the durums were the most sensitive cultivars. The most tolerant cultivars were either selections from local populations or had at least one parent of local origin. The detrimental effect of B on root dry matter production was much higher than on shoot dry matter (45 and 26%, respectively), but genotypical variation was greater in shoot growth retardation. While this implies the possible role of reduced translocation, high concentrations of B in the shoots of tolerant cultivars (though lower than in the sensitive cultivars) indicated the existence of other contributing mechanisms, such as tissue tolerance. Also, greater genotypical variation in older leaves showed that reduced uptake might be more important than reduced translocation in some cases. Due to the lack of correlation between results from the field and the controlled-environment studies, it was concluded that screenings should be undertaken in both situations as a means of verification. Another conclusion drawn was that symptom scoring for B tolerance was more reliable than measuring plant B concentrations.