For many globular proteins, the sequence and native structure are known. However, less is understood about how a string of amino acids folds into a functional protein. Experimental study of folding presents challenges due to the transience and variability of folding/unfolding transition states and intermediates. Alternatively, computational study of unfolding can provide significant insight into folding. Here, molecular dynamics simulations have been used to study the unfolding pathways of the SH3 domain structural family and to investigate the factors that determine the path and outcome. To separate folding determinants from amino acid sequence, 17 SH3 proteins were chosen with an average sequence identity of only 27%. Six unfolding simulations were performed for each protein, and the unfolding transition state ensemble was identified by locating the large, rapid conformational changes that signal the start of unfolding. Contact analysis was used to characterize the structure of the transition states ensembles. Two general pathways at the transition state were identified, distinguished based on the specific β-sheet structure lost at the transition state. In the first, more populated pathway contacts in the β-sheet containing the N- and C- terminal β-strands were lost while the second pathway was defined by structure loss in the other β-sheet. Though many of the investigated proteins went through both pathways in different simulations, most showed a clear bias towards one pathway. This work demonstrates that similar protein structures can fold through different pathways. The bias of many SH3 proteins towards one folding pathway also suggests the presence of some elements of primary structure that direct folding. Further investigation of the SH3 domain may yield ‘rules’ that determine the structure and folding pathway of the domain, and these rules may inform the study of other, similar proteins.