The tat trans-activator proteins of the primate immunodeficiency viruses contain a highly conserved cysteine-rich domain. In human immunodeficiency virus type 1 tat there are seven cysteines located between residues 22 and 37 that are thought to form a metal-nucleic acid-binding structure. Most of the previous mutagenesis studies had demonstrated that these residues are essential for tat activity and virus expression. Here we show that potentially conserved cysteine-histidine substitutions within the proposed tetrahedral structure still eliminate tat activity and virus expression. Consistent with previous studies, one cysteine-to-histidine mutation (amino acid 31) had little effect on trans-activation. We have studied the functional properties, stability and subcellular localization of several tat protein mutants. Most of the mutants are stable and properly localized to the nucleus and/or nucleolus. However, cysteine-to-glycine at position 34 affected tat stability. Our studies with the histidine mutants suggest that tat does not assume the prototype "zinc finger" structure for metal binding.