Obesity and associated metabolic disorders have become highly prevalent diseases worldwide, and the human gut microbiota, due to its influence on host energy metabolism, has been attributed an important role therein. This pilot study explores host-microbiota relationships in men and women affected by various types of glucose metabolism disorder. Among 20 individuals aged 58 to 71 years with either normal glucose tolerance, prediabetes, or type 2 diabetes mellitus the gut bacterial communities were compared based on barcoded 454 sequencing of 16S rRNA genes amplified from stool samples. We found that specific microbiota groups were relatively enriched or reduced in different metabolic states. Further, positive or negative associations with clinical manifestations of metabolic disease suggest that these organisms indicate and possibly contribute to metabolic impairment or health. For instance, a higher prevalence of Erysipelotrichaceae and Lachnospiraceae was found associated with metabolic disorders, and the Holdemania and Blautia genera correlated with clinical indicators of an impaired lipid and glucose metabolism. The Bacteroidetes and groups therein, by contrast, displayed inverse relationships with metabolic disease parameters and were found relatively enriched in participants not diagnosed with metabolic syndrome or obesity. Further, the prevalence of specific Clostridia and Rikenellaceae members also pointed towards a healthier metabolic state. Links with diet as an intermediate factor included positive and negative associations of Lachnospiraceae with relative consumption rates of fat and carbohydrates, respectively, and positive associations of Turicibacteraceae with the consumption of protein. Identifying critical roles of major gut microbiota components in metabolic disorders has important translational implications regarding the prevention and treatment of metabolic diseases by means of preventing or reversing dysbiosis and by controlling exacerbating diet and life style factors particularly in sensitive population groups.