I don't know if this counts as a correction, but in Pursuing Stacks Grothendieck mentions reading Quillen's book on model categories and was able to correct a mistake he made by reading a paper about model structures on the category of categories.