Milk and serum samples were collected from 16 cows naturally infected with M. paratuberculosis (detected by fecal culture; centrifugation method) and 9 uninfected cows (based on at least 3 negative annual fecal cultures). Samples were subjected to ELISA testing using 3 antigens: L-arbinomannan (LAM), protoplasmic (PPA), and protoplasmic following adsorpiton of M. phlei (Commonwealth Serum Laboratories [CSL]). Serum samples were tested with the LAM antigen within 24 hours of collection (fresh), and after being frozen (-70°C) for 1 year; only frozen samples were tested with PPA and CSL antigens. Fresh and frozen milk samples were evaluate with and without chemical preservative (Bronopal, brominated hydrocarbon) with the LAM antigen; only frozen samples were evaluated with PPA and CSL antigens. Milk ELISA scores were significantly correlated with serum ELISA scores (P less than 0.0001) for all 3 antigens, with r values ranging from 0.79 to 0.92. Repeated measures analysis of variance revealed no signficant difference between fresh vs. frozen milk or serum samples, nor was there any significant difference for preserved vs. unpreserved milk samples. Milk and serum ELISA scores were significantly higher for fecal culture-positive cows when compared with fecal culture-negative cows. However, there was considerable overlap in serum and milk ELISA scores for culture-positive and culture-negative cows. Apparently, freezing of serum and milk samples for 1 year at -70°C, and addition of chemical preservative to milk samples does not significantly affect the outcome of ELISA testing.