To determine if boron (B) deficiency, commonly reported to depress grain set in wheat, has the same effect in barley, a set of experiments compared five wheat and seven barley genotypes at various B levels in sand culture and in the field. In sand culture, plants were grown with levels of added B, from 0 to 10 muM. In the field, they were sown in a low B soil [0.15 mg hot water soluble (HWS) B kg(-1)] with three B treatments (nil, 2 t lime ha(-1), 1 kg B ha(-1)). In sand culture without added B, the genotypes ranged in grain set index (GSI) from 0 to 93% for wheat and 0 to 67% for barley. Boron concentration of the spike and flag leaf at booting in wheat and barley correlated (r = 0.8 - 0.9, P < 0.01) with the effect of B on GSI. Grain set was the only response, measurable in decreased number of grain spike(-1) and grains spikelet(-1), to low B in wheat. In barley, low B also depressed the number of spikelet spike(-1) by 23 to 75% and induced a 'rat-tail' symptom of terminal spikelet degeneration. There was a weak correlation between spike and flag leaf B and the effect of B on spike size in barley ( r = 0.47 and 0.37, respectively, P < 0.1). In some barley genotypes, the low B level that depressed grain set sometimes also delayed spike emergence and depressed the number of spikes plant(-1) but sometimes increased tillering and dry weight of straw. These results demonstrate that the phenotype of plant response to low B is more complex in barley than wheat and may require different strategies for managing B nutrition of barley including different approaches for selecting B efficient genotypes.