This study investigates dissolved, suspended loads and sands of major Icelandic rivers and determines chemical and mechanical erosion rates as well as rates of CO2 consumption by the chemical weathering. A steady state model of erosion is used to locally calculate the river suspended load fluxes needed to balance chemical weathering. The total dissolved solid concentrations range from 20 to 179 mg/kg. The highest concentrations are for spring fed rivers draining young rocks in the vicinity of active volcanoes, and the lowest for direct runoff rivers draining old Tertiary rocks. Total dissolved loads, "corrected" for atmospheric, geothermal, and magmatic inputs are used together with mean annual discharges to estimate low-temperature chemical erosion rates of 16 to 111 t/km2 /yr. These rates increase with runoff but decrease with the age of the rocks. Icelandic chemical erosion rates are higher than the world average for silicate rocks, reflecting both high reactivity of the basalt and high runoff, but lower than those for other basalt-draining rivers (in Réunion, Java, Azores or Deccan). CO2 consumption rates associated to chemical denudation range between 0.18 and 2.12 106 mol/km2/yr with an average value of 0.74 106 mol/km2/yr, higher than the world average for rivers draining silicate rocks. Chemical compositions of suspended sediments and sands are similar, showing a very low weathering stage. The elements most soluble during the weathering show slightly lower concentrations in the suspended sediments. River sediment chemical compositions are assumed to reflect a mixture between 3 initially pristine rock end-members: high Mg-basalt, tholeiite and rhyolite. The most insoluble elements (REE and Th) are used to re-define the mean chemical composition of the initially unaltered rocks of each drainage basin. A mass budget between the unaltered rock of the catchments and the river dissolved and suspended loads (steady state model of erosion) is used to calculate the average annual solid load of the rivers, which range from 650 to 4300 mg/1. For some rivers there is a good agreement between calculated and measured suspended loads but for others the calculated load is much higher than the measured one. The difference stems from groundwater inputs, man-made dams and other sedimentary traps. If the relevancy of the steady state model of erosion can be questioned, the accuracy of sediment load measurements is also questionable. Pros and cons of both methods are argued. The calculated solid loads lead to very high mechanical erosion rates, ranging from 940 to 10200 t/km2 /yr. Those increase with the glacier cover but decrease with the age of the catchment rocks. Icelandic mechanical erosion rates rank among the maximum reported rates, underscoring the importance of glaciers, tectonics, glassy basaltic rocks and high runoff. In association with low chemical weathering rates, these place Icelandic rivers as an end-member in the observed anti-correlation between mechanical to chemical erosion ratios and temperature for volcanic islands.