The processing of rod signals within the receptive field centre of cat retinal ganglion cells was investigated in two spot summation experiments by using the analytical methods of response and sensitivity summation. 2. The rod system was isolated by presenting test stimuli of short wave-length light against either a completely dark background or a dim background of long wave-length light. 3. Stimulus--response curves were obtained for two small, square-wave modulated test spots applied at points in the receptive field centre of equal sensitivity. The test spots were presented either singly or simultaneously. 4. In the absence of surround antagonism, the flux required to evoke a weak criterion response was the same whether the spots were presented singly or together. However, the flux required to evoke larger responses was typically half as great when the two spots were delivered together as it was when either was presented alone. 5. Over a moderate response range, the magnitude of response to the two test spots presented together equalled the algebraic sum of the two responses to the test spots presented alone. However, for responses of large magnitude, the algebraic sum was larger. 6. Permitting the surround to contribute substantially to the cell's response changed the outcome of the two spot summation experiment. 7. The data are consistent with a three stage model of signal processing within the receptive field centre: an early compressive power law transformation (within each sub-area) of illuminance into a neural signal which is followed by linear summation of sub-area signals and then a second compressive transformation.