We compared two methods for determination of serum zinc by flame atomic absorption spectroscopy. The first involved 10-fold dilution of serum with water; the second, deproteinization of serum with trichloroacetic acid. Analytical recovery of Zn2+ added to serum averaged 99.3 and 99.9%, respectively, by the two methods, but measured concentrations were consistently greater by the second method [mean difference, 118 +/- 27 (SD) mug/liter]. For the analysis of 55 sera, measured concentrations were also greater by the second method [mean difference, 91 +/- 28 (SD) mug/liter]. We dialyzed serum to remove most of the zinc and measured analytical recovery of increasing amounts of Zn2+ added to this dialyzed serum or, conversely, of zinc in a mixture of increasing volumes of dialyzed serum in a Zn2+ standard. Recovery results for the dilution method showed an interference by an unidentified, but nondialyzable serum component(s). Recovery was not improved by use of (a) a "high-solids" burner, (b) a diluent containing 1-butanol, or (c) a Mn2+ internal standard.