The development, life cycle, ultrastructure and phylogenetic position of an obligate, spore-forming endoparasite of Daphnia magna Straus is described. The microparasite was found in the body cavity of three Daphnia species (D. magna, D. pulex and D. longispina) collected in England and Russia during 1992-1994 and maintained in artificial culture by co-cultivation with D. magna. Transmission of the endoparasite occurred horizontally through waterborne spores released from the remains of dead infected hosts. Progeny of infected hosts were never infected, indicating that vertical transmission does not occur. Egg production by infected mothers ceased soon after infection and death ensued after 46 days (<latex>$\pm $</latex> 7 standard error) at 20 degrees C. Phase contrast light microscopy and transmission electron microscopy of the infection process showed the endoparasite to have a polymorphic life cycle beginning with the appearance of branched `cauliflower-like' rosettes and ended with the development of single, oval endospores, nippled at one end and with complex internal structure. Endospore formation resembled that found in endosporeforming bacteria. Morphologically the parasite has strong resemblance to the Pasteuria ramosa that Metchnikoff isolated from D. magna and D. pulex in Ukraine and described in 1888. Identification of this parasite has been an enduring puzzle since Metchnikoff. The previously confused phylogenetic position of P. ramosa (it has been classified as bacterium, yeast and protozoa) was resolved by sequencing the 16S rDNA molecule. Fluorescent in situ hybridizations confirmed that the 16S rDNA sequence obtained from the spores within the D. magna body cavity originated from the endoparasite. Maximum likelihood and maximum parsimony analysis showed that P. ramosa belongs to the low G+C Gram positive branch of the eubacteria and resides within a clade containing Bacillus tusciae, Alicyclobacillus cycloheptanicus and A. acidocaldarius as its nearest neighbours. These results confirm suggestions that this parasite is a bacterium and refute its previous tentative placement based on its morphological complexity among the Actinomycetales.