We have observed inaccurate urine arsenic values with the method of isobaric fractionation, which was designed to correct for the 40Ar35Cl interference with 75As quantitation by inductively coupled plasma mass spectrometry. Isobaric fractionation, which is based on ion intensities at m/z 77 and 82, consistently underestimates the 40Ar35Cl interference and overestimates urine arsenic. We present an improved method for identifying the argon-chloride interference. We observed that signal intensities for the species 16O35Cl and 40Ar35Cl are proportional (I75 = 0.0295 x I51 - 14.7, r2 = 0.998; where Ix is the normalized ion intensity at m/z X) in water and urine, over a broad range of chloride concentrations (0-800 mmol/L). The proportionality constant is remarkably stable within a run (mean and SD, 0.0295 +/- 0.0023, based on 10 replicates of five chloride calibrators, 0, 100, 200, 400, and 800 mmol/L). Increased sensitivity (50-fold) for detecting the 40Ar35Cl interference provides improved accuracy for urine arsenic quantitation as demonstrated by a split-sample comparison with graphite-furnace atomic absorption spectrophotometry.