The influence of sample preparation on measured concentrations of eight elements in the edible tissues of two black basses (Centrarchidae), two catfishes (Ictaluridae), and the black redhorse,Moxostoma duquesnei (Catostomidae) from two rivers in southeastern Missouri contaminated by mining and related activities was investigated. Concentrations of Pb, Cd, Cu, Zn, Fe, Mn, Ba, and Ca were measured in two skinless, boneless samples of axial muscle from individual fish prepared in a clean room. One sample (normally-processed) was removed from each fish with a knife in a manner typically used by investigators to process fish for elemental analysis and presumedly representative of methods employed by anglers when preparing fish for home consumption. A second sample (clean-processed) was then prepared from each normally-processed sample by cutting away all surface material with acid-cleaned instruments under ultraclean conditions. The samples were analyzed as a single group by atomic absorption spectrophotometry. Of the elements studied, only Pb regularly exceeded current guidelines for elemental contaminants in foods. Concentrations were high in black redhorse from contaminated sites, regardless of preparation method; for the other fishes, whether or not Pb guidelines were exceeded depended on preparation technique. Except for Mn and Ca, concentrations of all elements measured were significantly lower in cleanthan in normally-processed tissue samples. Absolute differences in measured concentrations between clean- and normally-processed samples were most evident for Pb and Ba in bass and catfish and for Cd and Zn in redhorse. Regardless of preparation method, concentrations of Pb, Ca, Mn, and Ba in individual fish were closely correlated; samples that were high or low in one of these four elements were correspondingly high or low in the other three. In contrast, correlations between Zn, Fe, and Cd occurred only in normallyprocessed samples, suggesting that these correlations resulted from high concentrations on the surfaces of some samples. Concentrations of Pb and Ba in edible tissues of fish from contaminated sites were highly correlated with Ca content, which was probably determined largely by the amount of tissue other than muscle in the sample because fish muscle contains relatively little Ca. Accordingly, variation within a group of similar samples can be reduced by normalizing Pb and Ba concentrations to a standard Ca concentration. When sample size (N) is large, this can be accomplished statistically by analysis of covariance; whenN is small, molar ratios of [Pb]/[Ca] and [Ba]/[Ca] can be computed. Without such adjustments, unrealistically large Ns are required to yield statistically reliable estimates of Pb concentrations in edible tissues. Investigators should acknowledge that reported concentrations of certain elements are only estimates, and that regardless of the care exercised during the collection, preparation, and analysis of samples, results should be interpreted with the awareness that contamination from external sources may have occurred.