Standard molecular simulation methods based on classical force fields typically assume only a fixed protonation state of systems. This assumption generally only permits a limited treatment of pH effects, for example, consideration of extreme acidic/basic conditions or situations where a small number of protonation states can be explicitly enumerated. Importantly, the standard approach cannot be scaled to chemical systems with a large number of titrateable sites such as lipid head groups or assemblies of protein subunits. Here we describe the development and application of a scalable and extensible method for including pH effects in molecular dynamics simulations. In contrast to other constant pH methods, the new method, based on a hybrid of non-equilibrium molecular dynamics and Monte Carlo, can be easily scaled to handle large heterogeneous systems, i.e., not only globular proteins. We present a validation of the method and its implementation in the program NAMD. Finally, we also present proof of concept studies on large biomolecular membrane systems.