The activity of -emitting radionuclides is usually measured by semiconductor detectors (surface barrier or ion implanted). Overlapping and composite bands are quite common problems depending on energy differences of the radionuclides and counting source preparation. The classical approach to activity quantification is based on peak integration and, when it is used, overlapping may be overcome by a detailed study of each case, whereas composite bands can not be completely resolved. Here, spectra of the -emitting plutonium isotopes, obtained by ion implanted semiconductor detectors, have been used to compare the classical approach with a multivariate calibration method (MVC-PLS). The study is performed at environmental activity levels (0–52 dpm). The relative errors obtained for239+240Pu activity determination, using either the classical or the MVC-PLS approach with replicates, are good enough to quantify isotopes at low level activities. The distribution of relative errors is asymmetric, with a positive component for 0–10.5 dpm subset, in the classical approach whereas it is more symmetric in the MVC-PLS method. The results show that the classical approach depends on peak overlap, whereas the MVC does not. As a whole, MVC is a more robust method than the classical approach. Composite bands were studied using the239Pu–240Pu mixture; the MVC approach did not allow individual quantification due to the lack of signal reproducibility. This instability does not affect the regular integration procedures but it is important in the deconvolution processes. The lack of reproducibility is related to the source preparation process.