Two methods, the multivariate linear regression and full spectrum technique in the fourier domain, were studied for simultaneous determination of cobalt(I), copper(I) and nickel(I) in mixture. Two programs, SPGRMLR and SPGRFSQ, were designed to perform these calculations. In the full spectrum method, seven error functions were calculated for estimating the number of factors and data reduction was fulfilled by factor analysis. Experimental results showed the full spectrum method to be more powerful than multivariate linear regression. Both methods were demonstrated to be successful in the present case.