All-atom molecular dynamics (MD) simulation is a valuable technique for providing detailed information about the dynamics of biomolecules, but its computational expense can often be prohibitive. This limitation has motivated the development of “enhanced sampling” simulation methods - purely algorithmic changes to conventional MD that aim to accelerate the sampling of configurational states. Although many such methods are available, relatively few systematic studies of their performance have been done. Quantitative claims about the performance of enhanced sampling simulations are typically limited to (i) comparisons with conventional MD simulations of small, model systems, which may present qualitatively different sampling challenges than complex biological systems, or (ii) comparisons with experimental data, which may be complicated by discrepancies between the actual experimental conditions and the modeling of those experimental conditions in the enhanced sampling simulation, including errors in the physical model, or force field, used in simulation. An effective alternative approach to quantifying performance of enhanced sampling methods would be to compare the results they obtain on complex biological systems directly to those obtained by conventional MD simulations using the same force field. Here, we use long-timescale conventional MD simulations to assess the performance of certain commonly used enhanced sampling methods in accelerating the sampling of protein conformational changes, protein folding, and ligand binding.