A theoretical model for nanofluid flow, including Brownian motion and thermophoresis, is developed and analysed. Standard boundary layer theory is used to evaluate the heat transfer coefficient near a flat surface. The model is almost identical to previous models for nanofluid flow which have predicted an increase in the heat transfer with increasing particle concentration. In contrast our work shows a marked decrease indicating that under the assumptions of the model (and similar ones) nanofluids do not enhance heat transfer. It is proposed that the discrepancy between our results and previous ones is due to a loose definition of the heat transfer coefficient and various ad hoc assumptions.