Radiometric (RCM) and conventional fecal culture (HEY) and a commercial polymerase chain reaction/DNA probe were evaluated as diagnostic tests for subclinical paratuberculosis in dairy cattle using fecal specimens from a repository of paratuberculosis specimens. The case definition of subclinical bovine paratuberculosis was isolation of Mycobacterium paratuberculosis, by conventional or radiometric culture, from fecal samples or internal organs of dairy cattle without diarrhea or chronic weight loss. Animals designated as free of the disease originated exclusively from certified paratuberculosis-free herds in Wisconsin. Among 182 infected cattle, RCM and HEY fecal culture and the DNA probe had test sensitivities of 54.4%, 45.1% and 33.5%, respectively. Fecal samples from only 111 of the M. paratuberculosis-infected cows tested positive by at least one of the three tests and these cows were designated as fecal shedders; the remaining 71 were considered to have prepatent infections. Among the 111 M. paratuberculosis fecal shedders, RCM, HEY and the probe detected the organism in 89.2%, 73.8% and 55.0% of the fecal specimens, respectively. Herd prevalence significantly affected the sensitivity of all three diagnostic tests (p less than 0.05) but only affected the fecal shedder detection efficiency of the DNA probe (p less than 0.01). No positive DNA probe results were found on 100 randomly selected fecal samples from cows in four certified paratuberculosis-free herds, thus the DNA probe was 100% specific. Probe analyses could be performed in 24 h or less. Time to complete the culture-based tests was 12 wk for HEY and 7 wk for RCM.(ABSTRACT TRUNCATED AT 250 WORDS)