Three field experiments were established to provide information upon which strategic decisions could be made on the advisability of breeding hybrid wheat versus pure lines for Australian wheat growing conditions. These experiments would determine the level and stability of heterosis for grain yield, provide data on how hybrids obtained their yield, indicate the accuracy of different planting arrangements for measuring heterosis and compare the performance of pure lines derived from heterotic hybrids with the F1 hybrids. Six hand-made, two produced using a chemical gametocide and two commercial hybrids were tested along with their parents (except for the commercial hybrids for which the parents were unavailable) in replicated hill plots in three environments (sowing date s) for two seasons, 1987 and 1988. The hybrids produced significantly more grain yield in all six environments and were marginally more stable than their parents. The best hybrid outyielded the commercial cultivar by 18.2 %. Mea n protein content of the hybrids was marginally lower r than that for the parents, i. e., 14.4 % compared with 14.8 %. Howe ver, the three high est yielding hybrids, despite their 14.5 —18. 3% yield advantage over the best yielding check cultivar were high er in protein than it by as much as 1.2% protein. Kernel weight appeared to contribute most consistently to yield heterosis. All hybrids were taller than the mid-parent value and half of them were taller than the high-parent. However, the increased height did not affect their harvest indices adversely. With only one exception the hybrids had significantly higher harvest indices compared with the mid-parent value. Eight of the hybrids from the above experiment were tested in hill plots, space-planted plots and in multiple row plots (large plots) planted at a seeding density comparable with commercial practice Genotype means from the three plot types for grain yield. Its components, plant height and harvest index were positively and significantly correlated. The three plot types determined genotypic differences with similar precision. Similar levels of heterosis were observed in hill plots and space-planted plots. 22.1 and 22.9%, respectively, whereas a relatively lower level was observed in large plots (11.7%). This reduction was mainly due to reduced heterosis for tiller number. The potential to develop pure lines with similar yield to the heterotic F1 hybrids was investigated using four hand-made hybrids and the commercial hybrid, Comet. Seventy five single seed descent derived F- (F6 in the case of Comet) lines were compared with their parents (except for Comet) and the F1 hybrid in replicated hill plots. Significant high-parent heterosis up to 15.1% for grain yield was observed for the hand-made hybrids. Estimation of heterosis for Comet was not possible because of the unavailability of its parents. All crosses produced pure lines which yielded similarly to their Fi hybrid indicating that pure lines of equivalent yield to the heterotic Fi could be developed by conventional breeding. There was no consistent pattern amongst the pure lines as to how they achieved their yield via their components. Optimal plant height was in the range of 70-90 cm, i.e., semidwarf types, for all crosses. The results of the experiments conducted indicate that although considerable and stable levels of heterosis were observed, pure lines of comparable yield to the hybrids could be produced. The decision to breed hybrids as opposed to breeding pure lines would appear therefore to depend on commercial rather than scientific considerations.