The effects of different methods on the distribution of total neutral sugars (TNS), uronic acids (UA), and beta-glucans (beta G) between the soluble (S) and insoluble (I) fractions of dietary fiber (DF) were determined for peas, kidney beans, oat bran, rice, and macaroni. Incorporation of a protease step into the Theander method "A" modestly increased, and addition of a pepsin digestion further increased the proportion of total fiber recovered in the S fraction. The effect of extraction method on the distribution of TNS, UA, and beta G between the S and I fractions varied with the food. The three methods measured the same total DF in a food and 1-3% starch in the I fraction of peas and kidney beans. Use of dimethyl sulfoxide to solubilize starch, or elevated temperature to extract S components, had no effect on the distribution of DF between S and I fractions of peas and macaroni. Incomplete protein hydrolysis did not always lower Klason lignin and excluding lignin from the fiber complex did not always substantially increase the S fraction.