The key technologies for cross-dispersed echelle spectrography,spectral reducing in real time and wavelength calibration, were researched.Based on the main dispersion, cross-dispersion and the relationship between them, the matrixes M1, M2 and M3 were established, and then the cross-dispersed echelle spectral matrixe Mλ-XY with the space coordinate on area CCD and the corresponding wavelength were given. By the characteristics of central wavelength and the free spectral range, the ideal data model without wavelength overlapping were obtained. A method to recognize the signal spots was proposed to locate the signal spot coordinate precisely. Using the data model without wavelength overlapping and the signal spot recognition, the two-dimensional overlapping echelle spectra were reduced and the wavelength was calibrated efficiently. By the result, this method is not only strong real-time in the echelle spectral analysis, but the wavelength accuracy is up to 0.01 nm. It can satisfy the requirements of echelle spectrograph for high resolutions, high precisions and rapid reading full spectra.