This work demonstrates the feasibility of the measurement of the isotopic composition of dissolved iron in seawater for an iron concentration range, 0.05-1 nmol L(-1), allowing measurements in most oceanic waters, including Fe depleted waters of high nutrient low chlorophyll areas. It presents a detailed description of our previously published protocol, with significant improvements on detection limit and blank contribution. Iron is preconcentrated using a nitriloacetic acid superflow resin and purified using an AG 1-x4 anion exchange resin. The isotopic ratios are measured with a multicollector-inductively coupled plasma mass spectrometer (MC-ICPMS) Neptune, coupled with a desolvator (Aridus II or Apex-Q), using a (57)Fe-(58)Fe double spike mass bias correction. A Monte Carlo test shows that optimum precision is obtained for a double spike composed of approximately 50% (57)Fe and 50% (58)Fe and a sample to double spike quantity ratio of approximately 1. Total procedural yield is 91 +/- 25% (2SD, n = 55) for sample sizes from 20 to 2 L. The procedural blank ranges from 1.4 to 1.1 ng, for sample sizes ranging from 20 to 2 L, respectively, which, converted into Fe concentrations, corresponds to blank contributions of 0.001 and 0.010 nmol L(-1), respectively. Measurement precision determined from replicate measurements of seawater samples and standard solutions is 0.08 per thousand (delta(56)Fe, 2SD). The precision is sufficient to clearly detect and quantify isotopic variations in the oceans, which so far have been observed to span 2.5 per thousand and thus opens new perspectives to elucidate the oceanic iron cycle.