A chemometrical method that allows one to obtain the isotopic abundances of the atoms composing a molecular ion, given its elemental composition and its mass spectrum, was developed. The method uses non-linear regression to fit the isotopic abundances of the elements, so that the experimental spectrum of the ion is reproduced. A least squares method with minimization by downhill simplex algorithm was implemented in C language as a library, but a graphical user interface application that allows an end-user to apply the method to his/her own spectra is also available. Examples of its application to electrospray mass spectra of NaAuCl 4 and CdCl 2 are given. The chlorine isotope pattern was obtained from the AuCl 4 − cluster in the former case and, simultaneously, chlorine and cadmium isotope patterns from the CdCl 3 − cluster. The performance of the method was extensively characterized by applying it to a great number of simulated spectra. Non-idealities, against which the method was evaluated, include noise, mass discrimination, detection non-linearity, base line offset (background counting), and low resolution. It has been demonstrated that, in spite of what intuition might tell one, the polyatomic approach to isotopic analyses has no a priori disadvantage to the common monatomic approach. Indeed, the results demonstrate that, in some cases, the figures of merit can be improved by using polyatomic species and that other polyisotopic elements can be included in the formula without significant loss of precision, as long as they have a dominant isotope, which is the case of carbon, hydrogen, oxygen, and nitrogen, for example. This means that polyatomic species derived from organic species could be useful for isotopic measurement purposes.