Five tests for antibodies against chlamydial (enzootic) abortion of ewes were compared using 255 sera from experimentally (group 1) or naturally (group 2) infected animals, flocks free of the disease (group 3) and individual animals testing positively by the complement fixation test but from flocks with no evidence of chlamydial abortion (group 4). Sera from five specific pathogen-free lambs vaccinated with two different subtypes of Chlamydia pecorum were also included (group 5). All tests used some form of processed culture of C psitiaci as antigen. Specificities, established with group 3 and 4 sera, ranged between 96 per cent (ELISA using lipopolysaccharide antigen) and 59 per cent (Immunocomb). Reactions with group 5 sera suggested that the cause of false positive results in the field might be cross-reactive antibodies against the arthritogenic subtype of C pecorum. Sensitivities, established with groups 1 and 2 sera, ranged between 81 per cent (Immunocomb) and 51 per cent (ELISA using solubilised protein antigen). The minimum sample sizes required to be 95 per cent certain of detecting at least five seropositives in two infected flocks (combined data) were 15 to 48, dependent on the test applied. The Western blot test, applied to a proportion of samples, yielded no false positives with group 3 sera but 31.7 per cent with group 4 sera. Thus, none of the tests in this comparison emerged as sufficiently satisfactory in all respects, suggesting that further improvements in chlamydial serology must come through the use of non-native antigens or in the form of a competitive ELISA.