Collisional damping of ion density fluctuations of an ion beam extracted from an ICP in a linear rf quadrupole cell [dynamic reaction cell (DRC)] is studied. Target gas pressure, parameter q, entrance and exit lens potentials and DRC rod offset were optimized to broaden the 107Ag+ ion packets of 0.2 ms duration to 2–4 ms (FWHM). It is shown that operating the DRC under such optimized conditions reduces the RSD of isotope ratio measurements obtained with a downstream quadrupole mass analyzer to the theoretical limit defined by the counting statistics error. For Ag and Pb samples at 40 ng ml−1, measured with the cell pressurized by Ne, isotope ratio RSDs of 0.02–0.03% at a counting statistics error of 0.01–0.02% are shown. All 4 isotopes of Pb were measured in the same acquisition. For NIST 982 Isotope Standard at a concentration of 10 ng ml-1, 204Pb/206Pb, 207Pb/206Pb and 208Pb/206Pb isotope ratios were determined with an external precision (n = 7) of 0.056–0.117%, 0.022–0.034% and 0.014–0.028% RSD, and with an average inaccuracy of 340 ppm, 781 ppm and 180 ppm, respectively. NIST 981 at 10 ng ml−1 was used as an external standard. Reactive elimination of isobaric interferences in the DRC is shown to allow precise isotope ratio measurement of normally interfered isotopes of Fe by a quadrupole ICP-MS. ArN+ and ArO+ interferences on Fe+ were removed by ion–molecule reactions with NH3. Collisional damping of the ion beam fluctuations was sustained and internal precision is shown to be limited only by counting statistics. For Fe at 50 ng ml−1, the average internal precision for 19 samples (9 replicates per sample) of 54Fe/56Fe and 57Fe/56Fe ratios was 0.068% and 0.1% RSD, with a corresponding average counting statistics error of 0.053% and 0.082% and external precision of 0.055% and 0.056% RSD, respectively.