Investigations were performed on the ZPPR-13A critical assembly to determine the cause of the radial variation of the calculated-to-experimental (C/E) ratio for control rod worth in large heterogeneous cores. The effects of errors in cross section, mesh size, group condensation, transport, and modeling were studied by using two- and three-dimensional diffusion calculations and three-dimensional transport calculations. In that process, the cross-section set and the calculation scheme that are being used for fast reactor design in India have been revalidated. The cross-section set was found to yield satisfactory results. Three-dimensional calculations with adjusted and unadjusted cross sections confirmed that the error in cross sections was largely responsible for the radial dependence of the C/E ratios. The contributions from group condensation and mesh size errors were <2%, and from modeling errors and transport correction, <1%. The effect of these errors is insignificant when compared with the effect of the cross-section error. The analysis also showed that even without the adjustment in diffusion coefficient suggested in earlier studies, a satisfactory prediction is found, at least for this benchmark. The diffusion-to-transport correction for control rod worth was found to be -7%.