In this work we present an alternative method for performing routine gamma spectra analysis based on genetic algorithm techniques. The main idea is to search for patterns of single nuclide spectra obtained by simulation in a sample spectrum targeted for analysis. We show how this approach is applied to the analysis of simulated and real target spectra, and also to the study of interference resolution.