The coating of nuclear fuel particles with pyrolytic carbon derived from a hydrocarbon gas is a complex process, and, until recently, although adequate behavior in service has been demonstrated, the methods used to obtain a particular product have been largely empirical. A concerted effort was made to close the loop: manufacture-quality-performance. A model of the decomposition process postulated the formation and growth of nuclei into agglomerates that are captured by the fuel particles. The evolution of the model involved many simplifications, and to reduce the number of variables involved, standardized operating conditions were assumed. The most important of these for comparative studies is the concept of operating at a constant reaction zone temperature. When this is done, many of the anomalies previously ascribed to the effect of different source gases are removed. An experimental program has been carried out to test the model, and excellent correlations have been found between the predicted and actual size of agglomerates that can be observed in the coating structure. The agglomerate size has also been correlated with coating failure. With the aid of the model, similar deposits have been made from quite different source gases. A survey of the failure modes of coated particle fuel acts as an aid to deducing, in the light of the deposition model, the method of achieving a satisfactory quality assurance program for the structure of coatings on nuclear fuel particles.