1. Particulate enzyme preparations obtained from Bacillus stearothermophilus B65 by digestion with lysozyme were shown to catalyse teichoic acid synthesis. With CDP-glycerol as sole substrate the preparations synthesized 1,3-poly(glycerol phosphate). It was characterized by alkaline hydrolysis, by glucosylation to the alkali-stable 2-glucosyl-1,3-poly(glycerol phosphate) with excess of UDP-glucose and a Bacillus subtilis Marburg enzyme system, by degradation of this latter product with 60%HF and periodate oxidation of the resulting glucosylglycerol. The specificity of the B. subtilis system previously reported (Glaser & Burger, 1964), was confirmed in the present work. 2. Pulse-labelling experiments, followed by periodate oxidation of the product and isolation of formaldehyde from the glycerol terminus of the polymer, showed that the B. stearothermophilus enzyme system transferred glycerol phosphate units to the glycerol end of the chain. The transfer reaction was irreversible. It was not determined if these poly(glycerol phosphate) chains were synthesized de novo, but it was shown that the newly synthesized oligomers were bound to much larger molecules. 3. When the B. stearothermophilus enzyme system was supplied with both CDP-glycerol and UDP-glucose, 1-glucosyl-2,3-poly(glycerol phosphate) was synthesized in addition to the 1,3-isomer. The former polymer was characterized by acid and alkaline hydrolysis, degradation with HF and periodate oxidation of the resulting glucosylglycerol, and periodate oxidation of the intact polymer followed by mild acid hydrolysis. This latter procedure removed the glucose substituents without disrupting the poly(glycerol phosphate) chain. 4. The poly(glycerol phosphate) isomers were distinguished by glucosylation with the B. subtilis enzymes and alkaline hydrolysis, the 2,3-isomer remaining alkali-labile. The proportion of 2,3-poly(glycerol phosphate) in the product increased with increasing amounts of UDP-glucose in the incubation mixture, but the total glycerol phosphate incorporated into products remained constant. It is suggested that the synthetic pathways of the two poly(glycerol phosphate) species may share a rate-limiting step.