Clonal bacterial cells can give rise to functionally heterogeneous subpopulations. This diversification is considered an adaptation strategy that has been demonstrated for several bacterial species, including Salmonella enterica serovar Typhimurium. In previous studies on mouse models infected orally with pure Salmonella cultures, derived bacterial cells collected from animal tissues were found to express heterogenous phenotypes. Here, we show mixed Salmonella populations, apparently derived from the same progenitor, present in human specimens collected at a single disease time point, and in a long-term-infected patient, these Salmonella were no longer expressing surface-exposed antigen epitopes by isolates collected at earlier days of the disease. The subpopulations express different phenotypes related to cell surface antigen expression, motility, biofilm formation, biochemical metabolism, and antibiotic resistance, which can all contribute to pathogenicity. Some of the phenotypes correlate with single nucleotide polymorphisms or other sequence changes in bacterial genomes. These genetic variations can alter synthesis of cell membrane-associated molecules such as lipopolysaccharides and lipoproteins, leading to changes in bacterial surface structure and function. This study demonstrates the limitation of Salmonella diagnostic methods that are based on a single-cell population which may not represent the heterogenous bacterial community in infected humans. IMPORTANCE In animal model systems, heterogenous Salmonella phenotypes were found previously to regulate bacterial infections. We describe in this communication that different Salmonella phenotypes also exist in infected humans at a single disease time point and that their phenotypic and molecular traits are associated with different aspects of pathogenicity. Notably, variation in genes encoding antibiotic resistance and two-component systems were observed from the subpopulations of a patient suffering from persistent salmonellosis. Therefore, clinical and public health interventions of the disease that are based on diagnosis of a single-cell population may miss other subpopulations that can cause residual human infections.