Abstract: Matrix transformation is often used in data processing. Ill-conditioning of matrix may result in the unstability of computational result. The condition number of matrix essentially reflects the morbidity problem. By means of the analysis of the condition number of matrix, the effect of matrix morbidity on data processing results for the fluorescence spectrum calibration is studied. Also experimental, results showed that the spectrum calibration matrix with a mutation data is more ill-conditioning and results in the distortion of the corresponding calibration, and data smoothing is necessitated to ensure high quality of the spectral calibration matrix.