Two field experiments were conducted to examine the relationships between tillering, spike number and grain yield in three winter wheat (Triticum aestivum L.) cultivars. Treatments were designed to manipulate both the production and survival of tillers, and to provide a high number of spikes per unit area. One experiment involved growth regulator treatments with cycocel and gibberellic acid while the second involved various rates of nitrogen. Grain yield was linearly related to total spike number over a range of 400–1200 spikes m−2 in a combined analysis over locations and years. When only spikes with at least nine fertile spikelets were included, a greater amount of the variability in yield could be explained, and differences among cultivars were related to the number of small, unproductive spikes. When locations and years were analyzed separately, little evidence was found for a diminishing response between grain yield and total spike number. Spike number was related to maximum tiller number in 1982, when winterkill and early spring conditions were unfavorable. Thus, although good fall tillering and winter survival contribute most to producing high spike numbers and grain yield, cultivars must also have the ability both to tiller rapidly in the spring and to sustain high-yielding tillers to provide sufficient compensation following winterkill.Key words: tillering, spikes, yield, wheat, nitrogen, regulators