We have developed a new model for coronavirus transcription, which we call discontinuous extension, to explain how subgenome-length negatives stands are derived directly from the genome. The current model called leader-primed transcription, which states that subgenomic mRNA is transcribed directly from genome-length negative-strands, cannot explain many of the recent experimental findings. For instance, subgenomic mRNAs are transcribed directly via transcription intermediates that contain subgenome-length negative-strand templates; however subgenomic mRNA does not appear to be copied directly into negative strands. In our model the subgenome-length negative strands would be derived using the genome as a template. After the polymerase had copied the 3'-end of the genome, it would detach at any one of the several intergenic sequences and reattach to the sequence immediately downstream of the leader sequence at the 5'-end of genome RNA. Base pairing between the 3'-end of the nascent subgenome-length negative strands, which would be complementary to the intergenic sequence at the end of the leader sequence at the 5'-end of genome, would serve to align the nascent negative strand to the genome and permit the completion of synthesis, i.e., discontinuous extension of the 3'-end of the negative strand. Thus, subgenome-length negative strands would arise by discontinuous synthesis, but of negative strands, not of positive strands as proposed originally by the leader-primed transcription model.