We have developed a highly precise method for the determination of ferrous iron (Fe 2+) in silicate rocks. Our new method is based on Wilson's procedure (1955) in which surplus V 5+ is used to oxidize Fe 2+ into Fe 3+ while equivalently reducing V 5+ into V 4+. Because V 4+ is more resistant to atmospheric oxidation than Fe 2+, Fe 2+ in the sample can be determined by measuring unreacted V 5+ by adding excess Fe 2+ after sample decomposition and then titrating the unreacted Fe 2+ with Cr 6+. With our method, which involves conditioning the sample solution with 5 M H 2SO 4 in a relatively small beaker (7 mL), the oxidation of Fe 2+ or V 4+ that leads to erroneous results can be completely avoided, even in 100-h sample decompositions at 100°C. We have measured the concentration of FeO in 15 standard silicate rock powders provided by the Geological Survey of Japan (GSJ). Analytical reproducibility was better than 0.5% (1σ) for all but those samples that had small amounts of Fe 2+ (<1.5 wt.% of FeO). Fourteen of these samples gave FeO contents significantly higher than the GSJ reference values. This likely indicates that the GSJ reference values, obtained by compiling previously published data, contain a large number of poor-quality data obtained by methods with lower recovery of Fe 2+ caused by oxidation or insufficient sample decomposition during analyses. To achieve accurate determinations of Fe 2+ in our method, several factors besides the oxidation must be considered, including: (1) long-term variations in the concentration of Fe 2+ solution must be corrected; (2) excess use of the indicator must be avoided; and (3) the formation of inert FeF + complex must be avoided during titration when using boric acid as a masking agent.