Bax is kept inactive in the cytosol by refolding its C-terminal transmembrane domain into the hydrophobic binding pocket. Although energetic calculations predicted this conformation to be stable, numerous Bax binding proteins were reported and suggested to further stabilize inactive Bax. Unfortunately, most of them have not been validated in a physiological context on the endogenous level. Here we use gel filtration analysis of the cytosol of primary and established cells to show that endogenous, inactive Bax runs 20-30 kDa higher than recombinant Bax, suggesting Bax dimerization or the binding of a small protein. Dimerization was excluded by a lack of interaction of differentially tagged Bax proteins and by comparing the sizes of dimerized recombinant Bax with cytosolic Bax on blue native gels. Surprisingly, when analyzing cytosolic Bax complexes by high sensitivity mass spectrometry after anti-Bax immunoprecipitation or consecutive purification by gel filtration and blue native gel electrophoresis, we detected only one protein, called p23 hsp90 co-chaperone, which consistently and specifically co-purified with Bax. However, this protein could not be validated as a crucial inhibitory Bax binding partner as its over- or underexpression did not show any apoptosis defects. By contrast, cytosolic Bax exhibits a slight molecular mass shift on SDS-PAGE as compared with recombinant Bax, which suggests a posttranslational modification and/or a structural difference between the two proteins. We propose that in most healthy cells, cytosolic endogenous Bax is a monomeric protein that does not necessarily need a binding partner to keep its pro-apoptotic activity in check.