Thirty sequenced microbial hydrogenases are classified into six classes according to sequence homologies, metal content and physiological function. The first class contains nine H2‐uptake membrane‐bound NiFe‐hydrogenases from eight aerobic, facultative anaerobic and anaerobic bacteria. The second comprises four periplasmic and two membrane‐bound H2 ‐uptake NiFe(Se)‐hydrogenases from sulphate‐reducing bacteria. The third consists of four periplasmic Fe‐hydrogenases from strict anaerobic bacteria. The fourth contains eight methyl‐viologen‐ (MV), factor F420‐ (F420) or NAD‐reducing soluble hydrogenases from methanobacteria and Alcaligenes eutrophus H16. The fifth is the H2‐producing labile hydrogenase isoenzyme 3 of Escherichia coli. The sixth class contains two soluble tritium‐exchange hydrogenases of cyanobacteria. The results of sequence comparison reveal that the 30 hydrogenases have evolved from at least three different ancestors. While those of class I, II, IV and V hydrogenases are homologous, i.e. sharing the same evolutionary origin, both class III and VI hydrogenases are neither related to each other nor to the other classes. Sequence comparison scores, hierarchical cluster structures and phylogenetic trees show that class II falls into two distinct clusters composed of NiFe‐ and NiFeSe‐hydrogenases, respectively. These results also reveal that class IV comprises three distinct clusters: MV‐reducing, F420‐reducing and NAD‐reducing hydrogenases. Specific signatures of the six classes of hydrogenases as well as some subclusters have been detected. Analyses of motif compositions indicate that all hydrogenases, except those of class VI, must contain some common motifs probably participating in the formation of hydrogen activation domains and electron transfer domains. The regions of hydrogen activation domains are highly conserved and can be divided into two categories. One corresponds to the ‘nickel active center’ of NiFe(Se)‐hydrogenases. It consists of two possible specific nickel‐binding motifs, RxCGxCxxxH and DPCxxCxxH, located at the N‐ and C‐termini of so‐called large subunits in the dimeric hydrogenases, respectively. The other is the H‐cluster of the Fe‐hydrogenases. It might comprise three motifs on the C‐terminal half of the large subunits. However, the motifs corresponding to the putative electron transfer domains, as well as their polypeptides chains, are poorly or even not at all conserved. They are present essentially on the small subunits in NiFe‐hydrogenases. Some of these motifs resemble the typical ferredoxin‐like Fe‐S cluster binding site. The variation of the sequences in these regions might determine the specific interaction between the external electron transfer domains of hydrogenases and their electron carriers. For example, the C‐terminal cysteine‐histidine rich region of the small subunits are distinguishable between those of H2‐uptake, H2‐producing, MV‐reducing, F420‐reducing and NAD‐reducing NiFe(Se)‐hydrogenases.