Identification of the genetic defect underlying early-onset diabetes is important for determining the specific diabetes subtype, which would then permit appropriate treatment and accurate assessment of recurrence risk in offspring. Given the extensive genetic and clinical heterogeneity of the disease, high-throughput sequencing might provide additional diagnostic potential when Sanger sequencing is ineffective. Our aim was to develop a targeted next-generation assay able to detect mutations in several genes involved in glucose metabolism. All 13 known MODY genes, genes identified from a genome-wide linkage study or genome-wide association studies as increasing the risk of type 2 diabetes and genes causing diabetes in animal models, were included in the custom panel. We selected a total of 102 genes by performing a targeting re-sequencing in 30 patients negative for mutations in the GCK, HNF1α, HNF4α, HNF1β and IPF1 genes at the Sanger sequencing analysis. Previously unidentified variants in the RFX6 gene were found in three patients and in two of them we also detected rare variants in WFS1 and ABCC8 genes. All patients showed a good therapeutic response to dipeptidyl peptidase-4 (DPP4) inhibitors. Our study reveals that next-generation sequencing provides a highly sensitive method for identification of variants in new causative genes of diabetes. This approach may help in understanding the molecular etiology of diabetes and in providing more personalized treatment for each genetic subtype.