A description is given of a γ-spectrum analysis method based on nonlinear least-squares fitting. The quality of the method is investigated by using statistical tests on the results from analyses of IAEA test spectra. By applying an empirical correction factor of 0.75 to the calculated peak-area uncertainties, agreement is achieved between the calculated and the observed variability of the results. The results in terms of relative peak areas and relative peak positions are shown to yield unbiased estimates of the tru values given by the IAEA. The overall accuracy of these estimates is significantly (P > 95%) better than that of estimates calculated from a previously reported analysis method. Standard deviations of the calculated peak-area ratios from the IAEA test spectra are given in order to permit the overall accuracy of other analysis methods to be compared with that of the present one.