Three blast furnaces were quenched in operation and the contents were examined; the temperature distribution was also measured, using Tempil pellets. The furnaces examined included a low productivity one, which was examined to see what was wrong. Changes in the quality of coke as it descends in the furnace, and coke behavior in the raceway and hearth are reported. The functions required of coke, and the effects of poor coke quality, are explained, together with the coke quality required in large blast furnaces. A theoretical study of the role of coke in large blast furnaces is included.