Natural products and their derivatives are main sources for lead compounds in drug discovery and development. Canonical natural product discovery relies largely on biological activity-guided or chromatographic identification-oriented screening strategies, which have achieved great success so far. However, the limitations of these methods, such as time consumption, labor intensity, and the noises of abundant natural products, have constrained productivities in discovering novel active natural products for drug development and combating the rising threat of drug resistance. Modern biotechnology, particularly the development of DNA sequencing and computational technology, has made it possible to study the biosynthesis of natural products, enabling us to connect genetic sequences with natural product structures for predicting the potentials of natural products produced by specific biological species at the genetic level. Therefore, genome mining-directed discovery for natural products has emerged. In addition to mining methods dependent on the conservation of genes encoding core enzymes for natural product biosynthesis, recently developed activity-oriented and intelligence-assisted genome mining strategies provide more opportunities for discovering naturally medicinal products. This article reviews the history of genome mining, highlighting advances in related databases, tools, and algorithms, with a focus on recent cases and applications of classic genome mining as well as self-resistance mechanism, evolutionary theory and artificial intelligence guided mining in the discovery of naturally active products. Since genomic information contains enormous chemical potentials, the discovery of natural products with high throughput and efficiency can accelerate the development of new drugs, new chemicals and new catalysts.