The US. Geological Survey operates a quality- assurance program based on the analyses of reference samples for the National Water Quality Laboratory in Arvada, Colorado, and the Quality of Water Service Unit in Ocala, Florida. Reference samples containing selected inorganic, nutrient, and low ionic-strength constituents are prepared and disguised as routine samples. The program goal is to determine precision and bias for as many analytical methods offered by the participating laboratories as possible. The samples typically are submitted at a rate of approximately 5 percent of the annual environmental sample load for each constituent. The samples are distributed to the laboratories throughout the year. Analytical data for these reference samples reflect the quality of environmental sample data produced by the laboratories because the samples are processed in the same manner for all steps from sample login through data release. The results are stored permanently in the National Water Data Storage and Retrieval System. During water year 1991, 86 analytical procedures were evaluated at the National Water Quality Laboratory and 37 analytical procedures were evaluated at the Quality of Water Service Unit. An overall evaluation of the inorganic (major ion and trace metal) constituent data for water year 1991 indicated analytical imprecision in the National Water Quality Laboratory for 5 of 67 analytical procedures: aluminum (whole-water recoverable, atomic emission spectrometric, direct-current plasma); calcium (atomic emission spectrometric, direct); fluoride (ion-exchange chromatographic); iron (whole-water recoverable, atomic absorption spectrometric, direct); and sulfate (ion-exchange chromatographic). The results for 11 of 67 analytical procedures had positive or negative bias during water year 1991. Analytical imprecision was indicated in the determination of two of the five National Water Quality Laboratory nutrient constituents: orthophosphate as phosphorus and phosphorus. A negative or positive bias condition was indicated in three of five nutrient constituents. There was acceptable precision and no indication of bias for the 14 low ionic-strength analytical procedures tested in the National Water Quality Laboratory program and for the 32 inorganic and 5 nutrient analytical procedures tested in the Quality of Water Service Unit during water year 1991.