Arsenic removal by passive treatment, in which naturally present Fe(II) is oxidized by aeration and the forming iron(III) (hydr)oxides precipitate with adsorbed arsenic, is the simplest conceivable water treatment option. However, competing anions and low iron concentrations often require additional iron. Application of Fe(II) instead of the usually applied Fe(III) is shown to be advantageous, as oxidation of Fe(II) by dissolved oxygen causes partial oxidation of As(III) and iron(III) (hydr)oxides formed from Fe(II) have higher sorption capacities. In simulated groundwater (8.2 mM HCO3(-), 2.5 mM Ca2+, 1.6 mM Mg2+, 30 mg/L Si, 3 mg/L P, 500 ppb As(III), or As(V), pH 7.0 +/- 0.1), addition of Fe(II) clearly leads to better As removal than Fe(III). Multiple additions of Fe(II) further improved the removal of As(II). A competitive coprecipitation model that considers As(III) oxidation explains the observed results and allows the estimation of arsenic removal under different conditions. Lowering 500 microg/L As(III) to below 50 microg/L As(tot) in filtered water required > 80 mg/L Fe(III), 50-55 mg/L Fe(II) in one single addition, and 20-25 mg/L in multiple additions. With As(V), 10-12 mg/L Fe(II) and 15-18 mg/L Fe(III) was required. In the absence of Si and P, removal efficiencies for Fe(II) and Fe(III) were similar: 30-40 mg/L was required for As(II), and 2.0-2.5 mg/L was required for As(V). In a field study with 22 tubewells in Bangladesh, passive treatment efficiently removed phosphate, but iron contents were generally too low for efficient arsenic removal.