The Multi-Filter Rotating Shadow-band Radiometer (MFRSR) makes precise simultaneous measurements of the direct solar beam extinction, and horizontal diffuse flux, at six wavelengths (nominally 415, 500, 615, 670, 870, and 940 nm) at one minute intervals throughout the day. MFRSR data obtained at tens of sites throughout the U.S. has been available for several years. This is a potentially very important but as yet underutilized data set. We describe and validate a retrieval algorithm for processing of MFRSR data from clear and partially cloudy days. This method uses consistency between direct normal and diffuse horizontal measurements together with a special regression technique to retrieve daily time series of column mean aerosol particle size, aerosol optical depth, NO2 and ozone amounts together with the instrument's calibration constants. Our validation studies demonstrate two advantages of our approach compared to the traditional Langley calibration method: less calibration variability and less sensitivity of retrievals to calibration accuracy. This method is currently used for processing data from a growing number of MFRSRs spread throughout the U.S. determining both time and geographic variability of aerosol properties and gaseous column amounts. This method makes the relatively inexpensive and automatic MFRSR an important tool in climatological research.