Following the detection of Johnes disease in sheep flocks in Victoria an eradication strategy was implemented on infected farms and an intensive tracing and surveillance program was initiated. In order to assess the performance of diagnostic tests for ovine Johne's disease specimens from 500 sheep from five farms were examined histologically and tested by agar gel immunodiffusion (AGID) and by absorbed EIA (PARACHEK®). The sheep were selected so that 80% were representative of the mob and 20% were from the 'tail' of the flocks with a body condition score < 2. Three of the flocks were selected on the basis of high prevalence of clinical disease and two with low prevalence of clinical disease. Infection was confirmed in 25% of the sheep with the prevalence in the 'tail' 2 to 8 times the prevalence in the mobs. The AGID had a sensitivity of 74% (37/50) for infected sheep in the 'tail' and overall sensitivity of 46.7% (56/120). To assess the specificity of PARACHEK® sera from 1000 Merino sheep from Western Australia which is not known to have Johne's disease were tested. Various cut-off algorithms were used to best determine the most appropriate for field use. By modifying the cut-off algorithm for cattle to negative control OD plus 0.200, the specificity was 99.3 ± 0.5%. Using the modified cut-off calculation PARACHEK® detected 44.2% (53/120) of infected sheep which is not significantly different from AGID. Both serological tests were effective in detecting infected sheep. In the mob with the lowest prevalence (2%) PARACHEK® was positive for one infected animal but the AGID was negative for both infected animals. Although all sheep were in poor condition due to limited feed following drought conditions, sheep with the lowest body score had a higher prevalence of Johne's disease. Sampling sheep with lowest body score enhances the sensitivity of serological detection of infection.