Diffusion modelling has shown that conservative demographic traits combined with feeding competition could explain red squirrel replacement by grey squirrels. We used field data from seven separate red and grey squirrel populations, in oak-hazel woods and Scots pines, to reject the hypothesis that red squirrel density and breeding is intrinsically poorer than that of grey squirrels. In oak-hazel woods, grey squirrel foraging, density and productivity were related to oak and acorn abundance. In contrast, red squirrels foraged where hazels were abundant; their relatively low density and breeding success were related to the abundance of hazel nuts. Red squirrels failed to exploit good acorn crops, although acorns were more abundant than hazels, but in Scots pines had densities and breeding success as high as grey squirrels in deciduous woods. Captive grey squirrels thrived on a diet of acorns, but red squirrels had a comparative digestive efficiency of only 59%, apparently because they were much less able than grey squirrels to neutralize acorn polyphenols. A model with simple competition for the autumn hazel crop, which was eaten by grey squirrels before the acorn crop, shows that red squirrels are unlikely to persist with grey squirrels in woods with more than 14% oak canopy. With oaks in most British deciduous woods giving grey squirrels a food refuge which red squirrels fail to exploit, replacement of red squirrels can be explained by feeding competition alone, exacerbated by the post-war decline in coppiced hazel.