In the last several years, our group has studied the folding behavior of several proteins, including single domain proteins such as cyt c, Rd-apocyt b562, barnase, PDZ domain, and FAT domain. We found that these proteins fold through partially unfolded intermediates that exist after the rate-limiting step. We called them """"""""hidden intermediates"""""""" since they can not be detected in conventional kinetic experiments. Further, we have developed a native-state hydrogen exchange-directed protein engineering method for populating the intermediates and determined their high-resolution structures by multi-dimensional NMR methods. Recently, we have extended our studies to include multi-domain proteins such as T4 lysozyme and ribonuclease H. The results obtained from these studies provide strong support for the hypothesis that the kinetic principle of protein folding is the step-wise folding of cooperative structure units (foldons) (see pictures in the Gallery). We also provided theoretical arguments on why proteins should fold in a step-wise manner and why the current funnel-like energy landscape view is inadequate to describe the folding behavior of proteins, i.e., desolvation during folding leads to energy barrier on the energy landscape. Proteins must do random search to fold through co-operative partially unfolded intermediates. We now have included the studies on the folding of intrinsically unfolded proteins when they are induced to fold by their partners.