Genomic selection (GS) is widely applied in tree breeding. Yet, population genetics theory indicates that GS could be further extended for predicting complex phenotypes in small-secluded natural populations, and guide their management and conservation. Furthermore, next-generation sequencing (NGS) provides the much-needed quantity of genetic markers, distributed throughout the genome, to allow estimating the proportion of phenotypic variance explained by genetic factors (heritability) without relying on structured pedigrees. This is at the base of GS models, which in tree breeding have been recurrently applied for improving commercial traits, yet not so often employed in natural stands. The aim of this chapter is to review the theoretical bases of GS and discuss how GS could be expanded into natural populations. After a meta-analysis of data found in the literature, we found a significant correlation between pedigree- (h2) and genome-(hg2) based heritabilities for key traits in forest and fruit trees, which allows the co-estimation of pedigree structure and heritability. Given that historical inbreeding in small-secluded populations usually increases relatedness and produces large, linked haplotype blocks, GS should be feasible in such stands. We discuss the challenges for performing GS predictions in the natural population and call for empirical studies to test such a possibility. GS should be a powerful tool to forecast adaptability to future environmental stressors (e.g., drought, heat, pests, and climate change overall) and assist management and conservation plans.