Ground-water sampling techniques were modified to reduce random low-level contamination during collection of filtered water samples for determination of trace-element concentrations. The modified sampling techniques were first used in New Jersey by the US Geological Survey in 1994 along with inductively coupled plasma-mass spectrometry (ICP-MS) analysis to determine the concentrations of 18 trace elements at the one microgram-per-liter (μg/L) level in the oxic water of the unconfined sand and gravel Kirkwood-Cohansey aquifer system. The revised technique tested included a combination of the following: collection of samples (1) with flow rates of about 2L per minute, (2) through acid-washed single-use disposable tubing and (3) a single-use disposable 0.45-μm pore size capsule filter, (4) contained within portable glove boxes, (5) in a dedicated clean sampling van, (6) only after turbidity stabilized at values less than 2 nephelometric turbidity units (NTU), when possible. Quality-assurance data, obtained from equipment blanks and split samples, indicated that trace element concentrations, with the exception of iron, chromium, aluminum, and zinc, measured in the samples collected in 1994 were not subject to random contamination at 1μg/L.Results from samples collected in 1994 were compared to those from samples collected in 1991 from the same 12 PVC-cased observation wells using the available sampling and analytical techniques at that time. Concentrations of copper, lead, manganese and zinc were statistically significantly lower in samples collected in 1994 than in 1991. Sampling techniques used in 1994 likely provided trace-element data that represented concentrations in the aquifer with less bias than data from 1991 when samples were collected without the same degree of attention to sample handling.