Wild ruminants are thought to serve as natural hosts for Rift Valley fever virus (RVFV) but the role of these animals as reservoirs for RVFV during inter-epidemic periods and as amplifiers during epidemics is not well understood. An indirect enzyme-linked immunoassay (I-ELISA) based on the recombinant nucleocapsid protein (rNp) of RVFV was validated for the detection of specific IgG antibodies in African buffalo. Data sets derived from testing buffalo sera from Kenya (n=405) and South Africa (n=618) were dichotomised according to the results of a virus neutralisation test. The assay characteristic performance was analysed using threshold values optimised by the two-graph receiver operating characteristics (TG-ROC) analysis, and by mean plus two, as well as by mean plus three standard deviations derived from I-ELISA PP values in uninfected animals. Among 1023 buffalo sera tested, 77 (7.5%) had detectable virus neutralising antibodies. The assay had high intra- and inter-plate repeatability in routine runs. At a cut-off optimised by the TG-ROC at 95% accuracy level, the diagnostic sensitivity of the I-ELISA was 98.7% and diagnostic specificity 99.36% while estimates for the Youden's index (J) and efficiency (Ef) were 0.98 and 99.31%. When cut-off values determined by traditional statistical approaches were used, the diagnostic sensitivity was 100% but estimates of J, Ef and other combined measures of diagnostic accuracy were lower compared to those based on cut-off value derived from the TG-ROC. Results of the study indicate that the I-ELISA based on the rNp would be useful for seroepidemiological studies of RVFV infections in African buffalo.