The use of response surface designs in preference to factorial designs in agriculture has become widely discussed and recommended. This paper simulates data for comparable response surface and factorial designs and uses this to demonstrate the similarities between the designs and their analyses and at the same time to point out some of the customary differences in their analyses. More particularly, it aims to show: (a) that for any design a reduction in the number of plots sown reduces the reliability of the results, and (b) that both a response surface design and a factorial design allow the calculation of estimated equations of response surfaces.