A three-stage interference detection and correction data processing scheme has been applied to Arecibo D region incoherent scatter power spectra. Individual power spectral estimates at a particular frequency are, in the absence of interference, Gaussian distributed about the mean for that frequency. The first processing stage involves rank ordering or median filtering techniques to estimate the parameters of the underlying Gaussian process at each spectral frequency. The second processing stage uses this information to construct a “window” which excludes the upper and lower 1% of the Gaussian distributed (interference free) power spectral data. Since interference power is additive, this window also excludes any interference contaminated spectral points that might also be present. The final processing stage involves weighted “fitting” of a model spectrum to the final corrected and averaged spectra. This three-stage signal processing scheme has yielded significant improvements in final data quality.