Water samples were collected during 1993 from 22 public supply wells screened in the Kirkwood-Cohansey aquifer system; concentrations of 18 trace elements were determined primarily by using inductively coupled plasma-mass spectrometry (ICP-MS) techniques, though graphite furnace atomic adsorption, hydride generation, and cold- vapor flameless atomic adsorption techniques were used for thallium, arsenic, and mercury, respectively, at the U.S. Geological Survey (USGS) National Water Quality Laboratory (NWQL). In addition, laboratory measurements of alkalinity and turbidity were made. The ground-water samples were collected by using ultra-clean sampling protocols developed by the USGS for collecting ground-water samples in areas with water containing low concentrations of trace elements. This technique is based on recently gained experience in sampling surface water for these elements. Field parameters (water temperature, specific conductance, pH, and dissolved-oxygen concentration) were monitored prior to sample collection. Three equipment blanks were collected to ensure that low-level trace-element contamination did not occur during sample collection. One split sample and a commercially- prepared reference standard were submitted to the NWQL o evaluate laboratory precision and accuracy, respectively. Trace-element concentrations in 10 sample splits and one equipment blank were also determined at the Rutgers University Chemistry Department laboratory. Results of the ICP-MS analyses and cold vapor flameless atomic absorption indicated that five trace elements-- cobalt, copper, lead, mercury, and nickel--were detectable in low concentrations (<0.1-29 mg/L) in most of the samples from the 22 wells, and four elements--aluminum, barium, manganese and zinc--were detected in higher concentrations than the other elements (30-710 mg/L for aluminum; 4-180 mg/L for barium, manganese, and zinc). The remaining nine trace elements were present in concentrations consistently lower than the minimum reporting limit. Turbidity was low (less than 1 nephelometric turbidity unit (NTU)), indicating that the trace-element concentrations were present in the dissolved phase and ideally would be reproducible in the absence of highly variable concentrations of particulates. The concentration of lead in one sample exceeded the U.S. Environmental Protection Agency (USEPA) action level of 15 mg/L; concentrations ranged from <1 to 16 mg/L. Mercury was frequently detected; concentrations ranged from <0.1 to 1.1 mg/L but did not exceed the USEPA maximum contaminant level. Results of analyses of the equipment blanks indicated that samples collected by using the new ultra-clean sampling protocols were free of low-level (< 1mg/L) trace-element contamination. The analysis of the split sample sent to the NWQL had a difference of 5 percent or less for all constituents except aluminum, for which the analysis had a difference of 10 percent. Results of ICP-MS analyses of split water samples sent to the Rutgers University Chemistry Department laboratory were, in general, in good agreement (within 10 percent) with those of the NWQL. Results of the analysis of the commercial standard agreed (within 5 percent) with the known concentrations of the trace elements. The quality-assurance data (three blanks, one split sample, and one standard), although not statistically evaluated because of the small data set, indicate that the measured trace-element concentrations are precise and accurate and that the samples were free of contamination at the microgram-per-liter level of contamination.