Bacterial 16S rDNA sequence data, incorporating sequences > 1 kb, were retrieved from published rumen library studies and public databases, then were combined and analysed to assess the diversity of the rumen microbial ecosystem as indicated by the pooled data. Low G+C Gram positive bacteria (54%) and the Cytophaga-Flexibacter-Bacteroides (40%) phyla were most abundantly represented. The diversity inferred by combining the datasets was much wider than inferred by individual studies, most likely due to different diets enriching for bacteria with different fermentative activities. A total of 341 operational taxonomic units (OTU) was predicted by the Chao1 non-parametric estimator approach. Phylogenetic and database analysis demonstrated that 89% of the diversity had greatest similarity to organisms which had not been cultivated, and that several sequences are likely to represent novel taxonomic groupings. Furthermore, of the 11% of the diversity represented by cultured isolates (> 95% 16S rDNA identity), not all of the bacteria were of ruminal origin. This study therefore reinforces the need to reconcile classical culture-based rumen microbiology with molecular ecological studies to determine the metabolic role of uncultivated species.