Optimal filters provide minimum variance interpolation and smoothing of noisy data, but the form of the noise power spectrum must be known for their design. We analyse data from the Microwave Limb Sounder (MLS) experiment onboard the Upper Atmosphere Research Satellite (UARS) to quantify parameters describing the noise spectra. MLS views space as part of its calibration sequence providing regular samples of the radiometer output. From these samples we determine breakpoint frequencies at which the spectral power density of the low frequency l/fequal to that of the white noise. Breakpoint frequencies were in the range from 0005 to 002Hz for the 90 filter channels of the MLS. A method was developed to distinguish between 11/fand systematic disturbances caused by cyclic changes in the operating environment as the spacecraft traverses its orbit.