Medium resolution (10-100m) optical sensor data such as those from the Landsat, SPOT, ASTER, CBERS and IRS-P6 satellites provide detailed spatial information for studies of ecosystems, vegetation biophysics, and land cover. While Landsat remains a cornerstone of medium resolution remote sensing, the ETM+ scan-line corrector failure in 2003 has highlighted the need for methods to integrate radiometry from multiple international sensors in order to create a consistent, long-term observational record. Such an approach needs to compensate for differing acquisition plans, sensor bandwidths, spatial resolution, and orbit coverage. Different processing approaches used in the calibration and atmosphere correction across sensors make integration even harder. In this paper, we propose a generalized reference-based approach to convert medium resolution satellite digital number (DN) to MODIS-like surface reflectance using MODIS products as a reference data set. This approach does not require explicit calibration and atmospheric correction procedures for individual medium resolution sensors, therefore minimizing the potential impact of those procedures due to among-sensor differences. Therefore, data in MODIS era from different sources such as Landsat TM/ETM+, IRS-P6 AWiFS, and TERRA ASTER can be combined for time-series analysis, biophysical parameter retrievals, and other downstream analysis. Our results from Landsat TM/ETM+ show that this approach can produce surface reflectance with a similar accuracy to physical approaches based on radiative transfer modeling with mean absolute differences of 0.0016 and 0.0105 for red and near infra-red bands respectively. The normalized MODIS-like surface reflectances from multiple sensors and acquisition dates are consistent and comparable both spatially and temporally with known trends in phenology.