Understanding the sequence elements coding for protein topologies is an important step toward determining structures from sequences. Here we use a Monte Carlo approach to generate the equilibrium conformations of the 56 amino acid B1 domain of protein G and several fragments, corresponding to progressive elongations from its N-terminus. This method, which is not biased towards any particular topology and starts from random conformations, yields alpha/beta topologies deviating by 3 Angstrom rms from the experimental structure. It is found that this simple protein model helps clarify the diversity of rate limiting steps that is observed in proteins sharing 15%-80% sequence identities with protein G.