ABSTRACT: Protein folding is a daunting problem. Difficulty resides in the extremely limited information provided by experiment. Protein domains fold into specific, structurally rich, three-dimensional patterns stabilized by myriads of weak interactions. In spite of such inherent complexity, protein domains appear to fold via a deceivingly simple two-state process in which individual molecules slowly alternate between the native 3D structure and an ensemble of unfolded conformations. Obtaining any mechanistic insight thus involves resolving the rare, but sudden, transitions between the native and unfolded states on single molecules with atomic detail and ultrafast time resolution. Meeting these three requirements simultaneously is inaccessible to experiment. For molecular dynamics simulations the limitation is how to access the long timescales required to observe folding and unfolding. However, the discovery of the downhill folding scenario drastically changes this state of affairs. This is so because downhill folding domains fold fast (in microseconds) via a gradual, structurally continuous, process. Therefore, downhill folding offers the opportunity of dissociating the three requirements. Our research efforts over the last decade have focused on developing methods to beat the structural, time, and single-molecule resolution limits on downhill folding proteins. In this presentation I will discuss some of these methods, our proof of principle results on a paradigmatic downhill folding domain, and how they can be generalized to other single-domain proteins. Finally, I will summarize some of the important findings that are emerging from our studies as well as their application for benchmarking the new large-scale molecular dynamics simulations capable of reaching the millisecond timescale.