Natural whey starters (NWS) are undefined bacterial communities produced daily from whey of the previous cheese-making round, by application of high temperature. As a result, in any dairy plant, NWS are continuously evolving, undefined mixtures of several strains and/or species of lactic acid bacteria, whose composition and performance strongly depend on the selective pressure acting during incubation. While NWS is critical to assure consistency to cheese-making process, little is known about the composition, functional features, and plant-to-plant fluctuations. Here, we integrated 16S rRNA metabarcoding and culture-dependent methods to profile bacterial communities of 10 NWS sampled in the production area of Parmigiano Reggiano cheese. 16S rRNA metabarcoding analysis revealed two main NWS community types, namely NWS type-H and NWS type-D. Lactobacillus helveticus was more abundant in NWS type-H, whilst Lactobacillus delbrueckii/St. thermophilus in NWS type-D, respectively. Based on the prediction of metagenome functions, NWS type-H samples were enriched in functional pathways related to galactose catabolism and purine metabolism, while NWS type-D in pathways related to aromatic and branched chain amino acid biosynthesis, which are flavor compound precursors. Culture-dependent approaches revealed low cultivability of individual colonies as axenic cultures and high genetic diversity in the pool of cultivable survivors. Co-culturing experiments showed that fermentative performance decreases by reducing the bacterial complexity of inoculum, suggesting that biotic interactions and cross-feeding relationships could take place in NWS communities, assuring phenotypic robustness. Even though our data cannot directly predict these ecological interactions, this study provides the basis for experiments targeted at understanding how selective regime affects composition, bacterial interaction, and fermentative performance in NWS.