To investigate how disulfide bonds can impact protein energy landscapes, we surveyed the effects of adding or removing a disulfide in two β‐lactamase enzymes, TEM‐1 and CTX‐M‐9. The homologs share a structure and 38% sequence identity, but only TEM‐1 contains a native disulfide bond. They also differ in thermodynamic stability and in the number of states populated at equilibrium: CTX‐M‐9 is two‐state whereas TEM‐1 has an additional intermediate state. We hypothesized the disulfide bond is the major underlying determinant for these observed differences in their energy landscapes. To test this, we removed the disulfide bridge from TEM‐1 and introduced a disulfide bridge at the same location in CTX‐M‐9. This modest change to sequence modulates the stabilities – and therefore populations – of TEM‐1's equilibrium states and, more surprisingly, creates a novel third state in CTX‐M‐9. Unlike TEM‐1's partially folded intermediate, this third state is a higher‐order oligomer with reduced cysteines that retains the native fold and is fully active. Sub‐denaturing concentrations of urea shifts the equilibrium to the monomeric form, allowing the disulfide bond to form. Interestingly, comparing the stability of the oxidized monomer with a variant lacking cysteines reveals the disulfide is neither stabilizing nor destabilizing in CTX‐M‐9, in contrast with the observed stabilization in TEM‐1. Thus, we can conclude that engineering disulfide bonds is not always an effective stabilization strategy even when analogous disulfides exist in more stable structural homologs. This study also illustrates how homo‐oligomerization can result from a small number of mutations, suggesting complex formation might be easily accessed during a protein family's evolution.This article is protected by copyright. All rights reserved.