Two atp6 genes were found in the mitochondrial genome of cytoplasmic male sterile (CMS) rice carrying the [cms-bo] cytoplasm. One (N-atp6) was identical to the normal cytoplasmic gene, while the second (B-atp6) was identified as a candidate CMS gene by Southern analysis of the mitochondrial genome of CMS cybrid rice. The coding sequence of B-atp6 was identical to the normal N-atp6 gene but its 3'-flanking sequence was different starting at 49 bases downstream from the stop codon. Northern analysis showed that B-atp6 is transcribed into a 2.0 kb RNA in the absence of the Rf-1 gene, whereas two discontinuous RNAs, of approximately 1.5 and 0.45 kb, were detected in the presence of the Rf-1 gene. Determination of the 3' and 5' ends of these RNAs suggested that the two discontinuous RNAs were generated from the 2.0 kb RNA by RNA processing at sites within the B-atp6-specific sequences by the action of the Rf-1 gene. Sequence analysis of cDNA clones derived from the N-atp6 RNA and the processed and unprocessed RNAs of B-atp6 indicated that the processed B-atp6 RNAs were edited as efficiently as the N-atp6, whereas unedited and partially edited RNAs were detected among unprocessed RNAs. RNA processing by Rf-1 thus influences the sequential post-transcriptional editing of the B-atp6 RNAs. Because the unprocessed RNAs of B-atp6 are possibly translated into altered polypeptides, our results suggest that interaction of RNA processing and editing plays a role in controlling CMS expression and the restoration of fertility in rice.