It is generally thought that, in order to compensate for lower solar flux and maintain liquid oceans on the early Earth, methane must have been an important greenhouse gas before similar to2.2 billion years (Gyr) ago(1-5). This is based upon a simple thermodynamic calculation that relates the absence of siderite (FeCO3) in some pre-2.2-Gyr palaeosols to atmospheric CO2 concentrations that would have been too low to have provided the necessary greenhouse effect(1). Using multi-dimensional thermodynamic analyses and geological evidence, we show here that the absence of siderite in palaeosols does not constrain atmospheric CO2 concentrations. Siderite is absent in many palaeosols (both pre- and post-2.2-Gyr in age) because the O-2 concentrations and pH conditions in well-aerated soils have favoured the formation of ferric (Fe3+)-rich minerals, such as goethite, rather than siderite. Siderite, however, has formed throughout geological history in subsurface environments, such as euxinic seas, where anaerobic organisms created H-2-rich conditions. The abundance of large, massive siderite-rich beds in pre-1.8-Gyr sedimentary sequences and their carbon isotope ratios indicate that the atmospheric CO2 concentration was more than 100 times greater than today, causing the rain and ocean waters to be more acidic than today. We therefore conclude that CO2 alone ( without a significant contribution from methane) could have provided the necessary greenhouse effect to maintain liquid oceans on the early Earth.