Protein engineering and kinetic experiments indicate that some regions of proteins have partially formed structure in the transition state for protein folding. A crucial question is whether there is a genuine single transition state that has interactions that are weakened in those regions or there are parallel pathways involving many transition states, some with the interactions fully formed and others with the structural elements fully unfolded. We describe a kinetic test to distinguish between these possibilities. The kinetics rule out those mechanisms that involve a mixture of fully formed or fully unfolded structures for regions of the barley chymotrypsin inhibitor 2 and barnase, and so those regions are genuinely only partially folded in the transition state. The implications for modeling of protein folding pathways are discussed.