We introduce a Hybrid Variable Flavor Number Scheme for heavy flavors, denoted H-VFNS, which incorporates the advantages of both the traditional Variable Flavor Number Scheme (VFNS) as well as the Fixed Flavor Number Scheme (FFNS). By including an explicit NF-dependence in both the Parton Distribution Functions (PDFs) and the strong coupling constant \alpha_S, we generate coexisting sets of PDFs and \alpha_S for NF={3,4,5,6} at any scale \mu, that are related analytically by the MS-bar matching conditions. The H-VFNS resums the heavy quark contributions and provides the freedom to choose the optimal NF for each particular data set. Thus, we can fit selected HERA data in a FFNS framework, while retaining the benefits of the VFNS to analyze LHC data at high scales. We illustrate how such a fit can be implemented for the case of both HERA and LHC data.