The Escherichia coli sigma 32 transcriptional regulator has been shown to be degraded both in vivo and in vitro by the FtsH (HflB) protease, a member of the AAA protein family. In our attempts to study this process in detail, we found that two sigma 32 mutants lacking 15-20 C-terminal amino acids had substantially increased half-lives in vivo or in vitro, compared with wild-type sigma 32. A truncated version of sigma 32, sigma 32 C delta, was purified to homogeneity and shown to be resistant to FtsH-dependent degradation in vitro, suggesting that FtsH initiates sigma 32 degradation from its extreme C-terminal region. Purified sigma 32 C delta interacted with the DnaK and DnaJ chaperone proteins in a fashion similar to that of wild-type sigma 32. However, in contrast to wild-type sigma 32, sigma 32 C delta was largely deficient in its in vivo and in vitro interaction with core RNA polymerase. As a consequence, the truncated sigma 32 protein was completely non-functional in vivo, even when overproduced. Furthermore, it is shown that wild-type sigma 32 is protected from degradation by FtsH when complexed to the RNA polymerase core, but sensitive to proteolysis when in complex with the DnaK chaperone machine. Our results are in agreement with the proposal that the capacity of the DnaK chaperone machine to autoregulate its own synthesis negatively is simply the result of its ability to sequester sigma 32 from RNA polymerase, thus making it accessible to degradation by the FtsH protease.