Pitting corrosion of carbon steel electrodes in 0.1 molL(-1) NaHCO3+0.02mol L-1 NaCl solutions was induced by anodic polarisation. The evolution of the breakdown potential E-b with NO2- concentration was investigated by linear voltammetry. E-b increased from - 15 +/- 5 mV/SCE for [NO2-1 = 0 up to 400 +/- 50 mV/SCE for [NO2-] = 0.1 mol L-1. During anodic polarisation at potentials comprised between E-b([NO2-] = 0) and E-b([NO2-] not equal 0), the behaviour of the whole electrode surface, followed by chronoamperometry, was compared to the behaviour of one single pit, followed via scanning vibrating electrode technique (SVET). Addition of a NaNO2 solution after the beginning of the polarisation led to a rapid repassivation of pre-existing well-grown pits. In situ micro-Raman spectroscopy was then used to identify the corrosion products forming inside the pits. The first species to be detected in the presence of NO2- were mainly dissolved Fe(III) species, more likely [Fe-III(H2O)(6)](3+) complexes. Iron(II) carbonate FeCO3, siderite, and carbonated green rust GR(CO32-) were also detected in the active pits, as in the absence of nitrite. But they were accompanied by maghemitey gamma-Fe2O3, a phase structurally similar to the passive film, that forms from the Fe(III) complexes. The Raman analyses then correlate with the SVET observations and confirm that the main effect of nitrite ions is to oxidize iron(II) into iron(III). The passive film would then form from the Fe(III) species still bound to the steel surface. (c) 2007 Elsevier Ltd. All rights reserved.