Concurrent bacteriologic culture of feces and agar gel immunodiffusion (AGID) testing was performed on all cows and bred heifers over 14 months old in 10 dairy herds during a 32-month period to determine the effectiveness of the AGID test for the detection of subclinical paratuberculosis. Herds were sampled 5 times and, when possible, culled animals were tested again at slaughter. During 5 herd-wide samplings, Mycobacterium paratuberculosis was isolated from 139 fecal specimens obtained from 109 cattle. Results of the AGID test were simultaneously positive 40 of 139 times (28.8%). Thirty-six of the 109 cattle (33.0%) determined to be infected had a positive AGID test result at some point during the 5 herd-wide samplings. When results of tests performed at time of slaughter were included, 117 cattle were identified as infected by culture methods; 55 of these (47.0%) were AGID test-positive at some point during the study. The upper limit of the maximal false-positive rate for the AGID test was 2.1%. On the basis of colony counts from cultures, subclinically infected cows shedding higher numbers of M paratuberculosis in their feces were more likely to have positive AGID test results (P less than 0.0001). In known infected cattle, neither the culture nor AGID test results were consistently positive on repeated testing. Of 48 official calfhood paratuberculosis vaccinates tested as adults, 3 had positive AGID test results and in 1 of these, M paratuberculosis was also isolated from the feces, indicating that the rate of false-positive AGID test results in calfhood vaccinates is low.