During service, parts made of spheroidal graphite (SG) cast iron may be submitted to temperature cycling. It is experimentally known that the growth of graphite can not be explained by the volume diffusion of carbon. Recent results obtained for the desaturation of the austenite in a SG cast iron confirmed the existence of a significant difference between the experiment and the theoretical predictions based on the volume diffusion of carbon. According to stress state calculations, one possible explanation for this discrepancy could rely on the existence of a strong hydrostatic component gradient close to the austenite / graphite interface that would act as a barrier for the carbon diffusion across this interface. However, this discrepancy seems not to be present in the case of the carbon diffusion to the free surface of the material when cycled under vacuum. In situ high temperature XRD has been used to follow the growth of the graphite layers on the free surface of the samples during cycling between 800 and 950degreesC. The results of these XRD tests are compared to previous results concerning the volume diffusion of carbon during graphitization.