Brazil, the largest Latin American country, is underrepresented in genomic research despite boasting the world's largest recently admixed population. In this study, we generated 2723 high-coverage whole-genome sequences from the Brazilian population, including urban, rural, and riverine communities representing diverse ethnic backgrounds. We reveal the impressive genomic diversity of Brazilians, identifying >
8 million previously unknown variants, including 36,637 predicted deleterious and potentially affecting population health. We found a positive correlation between these deleterious variants and ancestry. Brazilian genomes are a global haplotype mosaic shaped by nonrandom mating, with peak admixture in the 18th and 19th centuries. Within this diversity, ancestry-specific haplotypes exhibit an uneven spatiotemporal distribution. We also identified putatively selected genes in this diverse population, primarily linked to fertility, immune response, and metabolic traits.