First launched in 1972, the Landsat satellite sensors have provided the longest continuous record of high quality images of the Earth’s surface that are used in both civilian and military applications. Extraction of quantitative information (e.g., surface reflectance) from the Landsat image data is only possible through an accurate absolute radiometric calibration. Typically, this calibration has been performed as a radiance-based cross-calibration between sensors. However, to convert radiance to reflectance, an accurate estimate of solar exoatmospheric irradiance is critical; and there are several solar models currently available which estimate exoatmospheric irradiance with varying levels of accuracy. Because of these inconsistencies in solar models, a TOA reflectance-based approach, independent of exoatmospheric irradiance, has been developed to provide a consistent cross-calibration of the Landsat series (from Landsat 8 OLI to Landsat 4 MSS), based on analysis of coincident and near-coincident scene pairs acquired with each sensor. The methodology uses Landsat-8 OLI reflectance measurements as the starting point (reference), as they are estimated with a 3% uncertainty (compared to the 5% uncertainty associated with radiance measurements). A set of radiometric calibration coefficients has been estimated based on the equations presented in this paper, which allows direct conversion of the digital numbers from the image data to TOA reflectance. The results obtained from application of these coefficients show significant improvement in consistency of reflectance measurements among the Landsat sensors.