Data are presented for 45 Long Island groundwater samples each measured for 90Sr using four different analytical methods. 90Sr levels were first established by two New York State certified laboratories, one of which used the U.S. Environmental Protection Agency Radioactive Strontium in Drinking Water Method 905.0. Three of the 90Sr methods evaluated at Brookhaven National Laboratory can reduce analysis time by more than 50%. They were (a) an Environmental Measurements Laboratory Cerenkov technique and (b) two commercially available products that utilize strontium-specific crown-ethers supported on either a resin or membrane disk. Method independent inter-laboratory bias was <12% based on 90Sr results obtained using both U.S. Department of Energy/Environmental Measurements Laboratory and U.S. EPA/National Environmental Radiation Laboratory samples of known activity concentration. Brookhaven National Laboratory prepared a National Institute of Standards and Technology traceable 90Sr tap-water sample used to quantify test method biases. With gas proportional or liquid scintillation counting, minimum detectable levels (MDLs) of 37 Bq m(-3) (1 pCi L(-1)) were achievable for both crown-ether methods using a 1-L processed sample beta counted for 1 h. The primary radiological contaminants of the Brookhaven National Laboratory groundwater samples were 3H (14.8-518 kBq m(-3)) and 90Sr/Y (37 Bq m(-3)-18.5 kBq m(-3)). Thirty samples were above the U.S. EPA drinking water standard of 300 Bq m(-3) for beta emitting 90Sr. Regression of paired data sets showed that both crown-ether methods quantified 90Sr levels to within 5% of the "EPA method results" for samples greater than the drinking water standard and within 30-40% for samples less than the drinking water standard. The Cerenkov method, with a minimum detectable level of 150 Bq m(-3), was only useful for samples greater than the drinking water standard and underestimated 90Sr levels by 13%. Precision, as measured by the relative percent difference and mean difference statistics, was acceptable for both crown-ether methods when compared to the paired "EPA method" results. Cerenkov method precision exceeded pre-defined acceptance criteria for 12 of 33 samples analyzed.