The various kinetic and thermodynamic models for transcription elongation all require an understanding of the nature of the melted bubble which moves with the RNA polymerase active site. Is the general nature of the bubble system-dependent or are there common energetic requirements which constrain a bubble in any RNA polymerases? T7 RNA polymerase is one of the simplest RNA polymerases and is the system for which we have the highest-resolution structural information. However, there is no high-resolution information available for a stable elongation complex. In order to directly map melted regions of the DNA in a functionally paused elongation complex, we have introduced fluorescent probes site-specifically into the DNA. Like 2-aminopurine, which substitutes for adenine bases, the fluorescence intensity of the new probe, pyrrolo-dC, which substitutes for cytosine bases, is sensitive to its environment. Specifically, the fluorescence is quenched in duplex DNA relative to its fluorescence in single-stranded DNA, such that the probe provides direct information on local melting of the DNA. Placement of this new probe at specific positions in the non-template strand shows clearly that the elongation bubble extends about eight bases upstream of the pause site, while 2-aminopurine probes show that the elongation bubble extends only about one nucleotide downstream of the last base incorporated. The positioning of the active site very close to the downstream edge of the bubble is consistent with previous studies and with similar studies of the promoter-bound, pre-initiation complex. The results show clearly that the RNA:DNA hybrid can be no more than eight nucleotides in length, and characterization of different paused species suggests preliminarily that these dimensions are not sequence or position dependent. Finally, the results confirm that the ternary complex is not stable with short lengths of transcript, but persists for a substantial time when paused in the middle or at the (runoff) end of duplex DNA.