Analysis of 186Os/188Os variations in most natural samples is hampered by their low Os mass fractions (usually < 100 pg g−1), requiring Os separation from up to a kilogram of powder to obtain high precision measurement results. Recently developed amplifiers equipped with 1013 Ω feedback resistors improve the signal to noise ratio by an order of magnitude, allowing acquisition of high precision data (2SE < 50 ppm) from 186Os16O3− ion beams of ~ 40 fA and thus analysis of less material (~ 10–500 g). Factors controlling the uncertainty of 186Os/188Os N‐TIMS measurement results, using a mixed array of 1013 and 1011 Ω amplifiers, include isobaric interferences, baseline variability, gain calibration, decay time corrections and oxygen isotope composition variations. Isobaric interferences present the greatest challenges, but these can be largely overcome by using double Pt filaments and periodically monitoring interferences during each measurement. The optimised procedures produce 186Os/188Os ratios with precisions close to the theoretical limits imposed by counting statistics and Johnson–Nyquist electronic noise for ion beams of 10–70 fA. The intermediate measurement precision (85 ppm; 2s) approaches theoretically expected values (~ 50–80 ppm). This precision is sufficient to allow 186Os/188Os analyses of crustal rocks, for which relatively large variations are expected.