In this work, we present results of a tornado detection method applied to both simulated spectral data and real spectral data. Preliminary results with simulated spectral data obtained from a Rankine combined vortex reveal significant tornado spectral signatures, which deviate from conventional Gaussian-shaped weather Doppler spectra. For our real data tests, we used one of the few weather radars have the capability of collecting time series data (Level I data), the WSR-88D (KOUN) research radar at the National Severe Storm Laboratory (NSSL) in Norman, Oklahoma. Spectra from tornado outbreaks in Central Oklahoma on May 8 and 9, 2003 are presented. Flattened and non-Gaussian spectral feature are apparent in both cases. Results suggest that a valid tornado spectrum can be used to facilitate tornado detection.