Extensive comparisons of the measured dissolved organic carbon (DOC) and nitrogen (DON) contents of four different natural waters were carried out prior to the Seattle DOC/DON Workshop in July 1991 to assess the current variability in these measurements among aquatic scientists. This study included 34 DOC analyses with over 20 different instruments and 13 DON analyses of the same samples using nine different instruments. In spite of good precision for individual instruments (5.9% average sample mean deviation), independently measured DOC concentrations of the same water ranged over an order of magnitude and varied on average approximately +/- 40% of the mean value. Variability among similar instruments operated in different labs (+/- 15-42% of the mean) was almost as large. This observation plus evidence for constant offsets among DOC analyses made in different labs suggest that unaccounted for blanks contribute more toward the observed variability than other instrument characteristics such as oxidation method, catalyst or oven temperature. The corresponding relative analytical precisions for total dissolved nitrogen and DON were 6.8% and 27%, respectively. Independent measurements of DON varied on average +/- 30% of the mean value and could not be related to instrument characteristics within this highly variable data set. During the Workshop, manufacturers of four different DOC analyzers measured the DOC content of untreated and spiked lake and seawater samples using common blank and calibration procedures. Excellent agreement was obtained, supporting indications from the pre-workshop exercise that operator-dependent factors, such as blank determination and treatment, are more important sources of measurement variability than instrument design.