Two indirect ELISAs for the detection of antibodies against glycoprotein E (gE) of Aujeszky's disease virus (ADV) in sera have been developed. The rec-gE-ELISA is based on theE. coli-expressed recombinant protein containing the N-terminal sequences of gE (aa 1-125) fused with the glutathione S-transferase fromSchistosoma japonicum. The affi-gE-ELISA is based on native gE, which was purified from virions by affinity chromatography. The tests were optimised and compared with each other, as well as with the recently developed blocking gE-ELISA (Morenkov et al., 1997b), with respect to specificity and sensitivity. The rec-gE-ELISA was less sensitive in detecting ADV-infected animals than the affi-gE-ELISA (sensitivity 80% and 97%, respectively), which is probably due to the lack of conformation-dependent immunodominant epitopes on the recombinant protein expressed inE. coli. The specificity of the rec-gE-ELISA and affi-gE-ELISA was rather moderate (90% and 94%, respectively) because it was necessary to set such cut-off values in the tests that provided a maximum level of sensitivity, which obviously increased the incidence of false positive reactions. Though the indirect ELISAs detect antibodies against many epitopes of gE, the blocking gE-ELISA, which detects antibodies against only one immunodominant epitope of gE, showed a better test performance (specificity 99% and sensitivity 98%). This is most probably due to rather high dilutions of the sera used in the indirect gE-ELISAs (1:30) as compared to the serum dilution in the blocking gE-ELISA (1:2). We conclude that the indirect gE-ELISAs are sufficiently specific and sensitive to distinguish ADV-infected swine from those vaccinated with gE-negative vaccine and can be useful, in particularly affi-gE-ELISA, as additional tests for the detection of antibodies to gE.