The amino acid sequences of the B800/850 light-harvesting proteins from Rhodospirillum molischianum were determined by Edman degradation. On the basis of these amino acid sequences, two degenerated oligonucleotides were synthesized and used for PCR of genomic DNA. The resulting 150 bp DNA fragment was cloned, sequenced and used for subsequent Southern blot analysis of digested genomic DNA. A 2.3 kbp EcoRI fragment strongly hybridized to the probe and a size selected genomic library from genomic DNA was constructed. One clone scored positive during screening of the library with the PCR-fragment and subsequent DNA sequence analysis of the clone revealed the presence of three A-genes (A1A2A3) encoding alpha-polypeptides and of two B-genes (B1B2) encoding beta-polypeptides of the B800/850 complex. The arrangement of the different genes are B1A1, B2A2 and A3 where only B1 and B2 are preceded by typical Shine-Dalgarno sequences. In addition, typical nucleotide sequences for a rho-independent termination of transcription are located downstream of the genes A1 and A2. The deduced amino acid sequences revealed that the alpha-genes encoded for identical polypeptides, whereas the deduced beta-polypeptides differed in their amino acid sequence at four positions. Transcriptional operon analysis revealed that the genes A1B1 and A2B2 are both dicistronically transcribed, whereas the gene A3 is not.