This paper focuses on the "spectral" aliasing phenomenon that may produce distortions on remotely sensed spectra acquired by hyper-spectral push-broom sensors and that arises because of an inadequate sampling rate. The analysis of the aliasing appearance has been performed on a set of at-sensor radiance spectra computed stemming from some spectral libraries with spectral resolution sufficiently high for our aims. A general procedure to evaluate aliasing in spectral remote sensing data has been proposed. A model for the system modulation transfer function of a hyper-spectral push-broom sensor (like PRISM) has been developed by taking into account the different contributions due to optics, electronics, detector, spectrometer dispersion and satellite motion. By using this sensor model, the set of high resolution spectra has been processed in order to obtain the related set of simulated acquired spectra; also a corresponding set of not aliased spectra (i.e. the spectra that would be produced if the Nyquist condition would be fulfilled) has been produced. Several score indexes have been considered among those proposed in literature and the three most effective have been implemented and applied to evaluate the aliasing produced in the acquired data by comparing the aliased and not aliased spectra. Aliasing evaluation has been first performed onto the simulated spectra without atmospheric and radiometric correction. Afterwards "ideal" correction based on the knowledge of the ground irradiance and the atmospheric transmittance spectrum has been implemented, hence the aliasing evaluation has been performed also onto the reconstructed set of spectra (atmospherically and radiometrically corrected). Results are presented in the paper. To remove the constraints of the "ideal correction", a simple atmospheric correction model has been implemented and applied to the simulated spectra radiometrically corrected; a qualitative evaluation of the reconstructed spectra has been performed.