ABSTRACT: Marine microbial diversity is important yet poorly-known, due to low culturability and undersampling. However, 16S rRNA gene sequences cloned directly from biomass allow us to know what microbial types are present, irrespective of culturing, and to createprobes suitable for biodiversity studies. Many sequences are needed for good probe design. Here we report on sequences from 57 deep sea clones, obtained by the polymerase chain reaction with 'universal' primers, from 500 m and 3000 m depths in thenortheast Pacific and 1000 m depth in the subtropical Atlantic. The most common group, with 19 of the new sequences (10 Atlantic), was a recently reported crenarchaeal cluster, Group I. We also found 6 sequences in 2 other archaeal groups in the broadmethanogen-halophile lineage; 2 of these were in a distinct lineage not previously reported. The bacterial sequences included 22 dispersed among the alpha and gamma Proteobacteria (8 related to SAR 11), 5 related to a previously reported broad group(Group A) of marine clones poorly affiliated with known (cultured and sequenced) major bacterial divisions, 6 in a second group with little affiliation to any previously reported division (we call this Group B), 1 in a third possible major novel group, 2deeply branched within the 'Green Nonsulfur' lineage, and 1 branching with a soil clone. In contrast to the vast majority of the sequences, a cluster of 5 sequences was very close to a cultured marine proteobacterium, Alteromonas macleodii. Itappeared that 5 of the clones were chimeric, although this label is difficult to apply when sequences are only distantly related to those in the database, as was common. We conclude that the deep sea contains numerous novel and widespread majorprokaryotic lineages. Given the huge volume of this habitat and typical bacterial abundances, it appears that the previously unknown archaeal and bacterial groups may be the most abundant organisms on Earth.