The early identification of children with learning difficulties is examined critically. It is concluded that the status of this process is in doubt both at a conceptual level, and in terms of the research evidence. The various instruments used are shown to be of varying usefulness, but even the better ones are open to criticism. The argument presented is that this process needs to be interactionist in its widest sense. That is, it is important to consider not simply the child's own abilities, but the external effects, including social factors. Also, the philosophy of the educational institution is considered important as the aims therein, implicit and explicit, will affect not only how children are taught but how learning difficulties are conceptualised.