Winter wheat cultivars Apollo, Hornet, Longbow and Norman were each sown at 50, 100, 200, 400, 800 and 1600 seeds/m2 in a field experiment conducted in Northern Ireland over the 1989/90 crop year. No growth regulators were applied and the wheat received 178 kgN/ha top-dressing in the spring. Hourly rainfall, windspeed and wind direction data were recorded and lodging was visually assessed from the end of May to harvest. Lodging first occurred in the 1600 seed-rate plots as the ears were emerging in early June and then progressively increased in the 800, 400 and 200 plots during June, July and August. Lodging did not occur suddenly but took several hours, with the stems first lying at an angle before lodging completely. Stem buckling or breakage did not appear to be the principal form of structural failure. The longest strawed cultivar, Longbow, lodged most severely. The shorter-strawed Norman also lodged badly in contrast with Hornet which had a similar straw length. Apollo, which was taller than Norman and Hornet and produced more ears per square metre than the other cultivars, lodged least but tended to lean at c. 30° from the vertical. Lodging occurred during or within 24 h of periods of rainfall which, in many cases, coincided with windspeeds at crop height averaging > 25 km/h and occasionally > 50 km/h. Lodging also occurred following rainfall when the windspeed did not exceed 16 km/h. The grain yield was negatively correlated with the average lodging from ear emergence to harvest, there being a 1 t/ha decline in yield for each 10% increase in average area lodged. The 50 and 100 seed-rate plots yielded 10 t/ha and had little or no lodging. The decline in yield with increased lodging and seed rate was attributed to the effect of lodging rather than to seed rate and was associated with a fall in the number of grains/ear and 1000-grain weight from 56 and 53·5 g at the lowest seed rate to 15 and 42·7 g at the highest, respectively. A comparison of the plants from lodged and unlodged plots of the 1600 and 800 seed rates, and subsequently of the 800 and 400 seed rates, indicated that at the higher seed rate, lodged plots had less fresh weight per unit area, basal internodes with smaller diameters, fewer support roots per stem, and a lower root dry weight per stem.