Many bacterial genomes encode multiple metal-sensing ArsR-SmtB transcriptional repressors. There is interest in understanding and predicting their metal specificities. Here we analyse two arsR-smtB genes, ydeT and yozA (now aseR and czrA) from Bacillus subtilis. Purified AseR and CzrA formed complexes in gel-retardation and fluorescence-anisotropy assays with fragments of promoters that were derepressed in ΔaseR and ΔczrA cells. Candidate (i) partly thiolate, α3-helix (for AseR) and (ii) tetrahedral, non-thiolate, α5-helix (for CzrA) metal binding sites were predicted then tested in vitro and/or in vivo. The precedents are for such sites to sense arsenite/antimonite (α3) and zinc (α5). This correlated with the respective metal inducers of AseR and CzrA repressed promoters in B. subtilis and matched the metals that impaired formation of protein–DNA complexes in vitro. The putative sensory sites of 1024 ArsR-SmtB homologues are reported. Although AseR did not sense zinc in vivo, it bound zinc in vitro exploiting α3 thiols, but AseR DNA binding was not impaired by zinc. If selectivity relies on discriminatory triggering of allostery not just selective metal binding, then tight non-effector metal complexes could theoretically inhibit metal sensing. AseR remained arsenite-sensitive in equimolar zinc, while CzrA remained zinc-sensitive in equimolar arsenite in vitro. However, cupric ions did not impair CzrA–DNA complex formation but did inhibit zinc-mediated allostery in vitro and prevent zinc binding. Access to copper must be controlled in vivo to avoid formation of cupric CzrA.