The accuracy of an immunoassay for mucin depends on the antibodies' specificity. During human small intestinal mucin purification on a CsCl gradient, a very antigenic non-mucin contaminant was found at 1.48 g.ml-1 density. This was separated from the mucin (1.42 g.ml-1 density) by dividing the gradient into 26 fractions, but not by dividing it into 8 fractions. Polyclonal antibodies raised against mucin obtained using 8 fractions reacted with mucin and contaminant, but antibodies raised against mucin obtained using 26 fractions reacted only with mucin. The identity of the contaminant is unknown. However, it contained nearly equal amounts of galactose, N-acetylglucosamine and glucose, but did not react strongly with periodate-Schiff reagent. Immunofluorescence microscopy showed it was located in the small intestinal goblet cells and mucus layer. Unless the contaminant is removed before raising polyclonal antibodies to mucin, immunoassays will give inaccurate results.