Convincing concepts of redox control of gene transcription have been worked out for prokaryotes and lower eukaryotes, whereas the knowledge on complex mammalian systems still resembles a patchwork of poorly connected findings. The article, therefore, reviews principles of redox regulation with special emphasis on chemical feasibility, kinetic requirements, specificity, and physiological context, taking well investigated mammalian transcription factor systems, nuclear transcription factor of bone marrow-derived lymphocytes (NF-κB), and kelch-like ECH-associated protein-1 (Keap1)/Nrf2, as paradigms. Major conclusions are that (i) direct signaling by free radicals is restricted to O(2)•- and •NO and can be excluded for fast reacting radicals such as •OH, •OR, or Cl•; (ii) oxidant signals are H(2)O(2), enzymatically generated lipid hydroperoxides, and peroxynitrite; (iii) free radical damage is sensed via generation of Michael acceptors; (iv) protein thiol oxidation/alkylation is the prominent mechanism to modulate function; (v) redox sensors must be thiol peroxidases by themselves or proteins with similarly reactive cysteine or selenocysteine (Sec) residues to kinetically compete with glutathione peroxidase (GPx)- and peroxiredoxin (Prx)-type peroxidases or glutathione-S-transferases, respectively, a postulate that still has to be verified for putative mammalian sensors. S-transferases and Prxs are considered for system complementation. The impact of NF-κB and Nrf2 on hormesis, management of inflammatory diseases, and cancer prevention is critically discussed.