Reaction centers from wild-type Rhodobacter sphaeroides (formerly called Rhodopseudomonas sphaeroides) were separated into two components: the LM complex and H subunit. LM was isolated after brief treatment of reaction centers with SDS by affinity chromatography with cytochrome c as ligand. A stable H preparation was obtained after dissociation of reaction centers with lithium perchlorate. LM was depleted of the transition metal, Mn, which interacts with QA and QB in native reaction centers. It retained only 30% of primary photochemistry which could be restored to 50-80% by addition of Q6, Q10 or other quinones. A stable semiquinone radical QA- could be flash-induced in LM. Its absorption properties are similar to those of QA- in native reaction centers. The quantum yield of photochemistry in an LM unit reconstituted with Q6 is the same as in intact reaction center and in LM in the presence of H. This result was confirmed by the rapid electron-transfer rate between I- and QA in LM + H (.tau. .apprxeq. 0.45 ns). Ubiquinone in LM incubated with H becomes tightly bound at the QA site. Flash production of a Q2- species was not detected in LM and LM + H. We conclude that the depletion of the reaction center both of the H subunit and of the metal does not necessarily lower the quantum yield of the primary reaction or greatly modify the rate of electron transfer from I- to QA. These results contrast with observations of others that seemed to demonstrate that the metal is essential for high-rate electron transfer between I- and QA (Debus, R.J., Feher, G. and Okamura, M.Y. (1986) Biochemistry 25, 2276-2287). In our experiments, secondary electron transfer to QB was not restored in LM + H, unlike in reconstitution experiments reported with R26 Rb. sphaeroides reaction centers (Debus, R.J., Feher, G. and Okamura, M.Y. (1985) Biochemistry 24, 2488-2500). Apparently, interactions between H and LM were too weak for restoring QB activity.