The high variability that exists in the NIH potency test for inactivated rabies vaccines is well known. To control this variability, the use of a reference preparation of rabies vaccine in parallel with the test vaccines is recommended. However, our data indicate, that the NIH test is more variable when calculating antigenic values using the reference vaccine CRV1 than when calculating only PD50 values. We immunized the mice with a single dose of different vaccine dilutions. As our results suggest, the antibody assay in immunized mice used for the NIH test, seems to be the best possible manner to determine the potency of inactivated rabies vaccines. The antigenic value of the NIH test does not correlate with the antibody status in immunized cats and dogs, but a correlation between antibody titres in mice and dogs appears to exist.