Fusarium head blight (FHB) affects whole spikes of small grain plants, yet little is known about how FHB develops following infection, or about the concentration or progression of the mycotoxin deoxynivalenol (DON) in non-grain spike tissues. Fusarium mycotoxin levels in whole small-grain spikes are of concern to producers of whole-crop silage, as well as users of straw containing chaff for animal bedding or winter livestock rations. A 2-year field experiment was performed in Kinston, NC to reveal the time course of FHB development. Eight winter wheat cultivars with varying levels of FHB resistance were used in the 2006 experiment, and four of them were used in 2007. Plots were spray-inoculated with Fusarium graminearum macroconidia at mid-anthesis. Four durations of post-anthesis mist were applied: 0, 10, 20, or 30 days. Spike samples were collected and bulked by plot at 15, 25, 35, 45, 55, and 65 days after anthesis (daa); samples were separated into grain, glume, and rachis fractions. Increasing durations of post-anthesis moisture elevated grain DON and reduced the effect of cultivar on DON, presumably by affecting the expression of resistance, in all spike tissues. Fusarium-damaged kernels increased from early kernel-hard to harvest-ripe in both years. Percent infected kernels increased from medium-milk to harvest-ripe. During grainfill, DON concentrations declined in grain but increased in rachises and glumes, peaking at early kernel-hard, before declining. Higher mean and maximum DON levels were observed in rachises and glumes than in grain. Estimated whole-spike DON peaked at early kernel-hard. In a high-FHB year, whole-plant harvest for forage should be conducted as early as possible. Straw that may be consumed by livestock could contain significant amounts of DON in chaff, and DON can be minimized if straw is sourced from low-symptom crops. Cultivar FHB resistance ratings and disease data should be useful in predicting whole-spike DON levels. Overall, associations between grain DON levels in harvest-ripe and prior samples were stronger the later the prior samples were collected, suggesting limits to the possibility of predicting harvest-ripe grain DON from earlier levels.