We present recent work that applies smFRET techniques to the early stages of bacterial transcription in which DNA is transcribed into RNA, resulting in a FRET change as a DNA template is transcribed. The mechanism of initial transcription is unclear, mainly due to challenges arising from transient intermediates and molecular heterogeneity. Our work has revealed that RNA polymerase pauses after producing a 6-mer RNA, and that the pause can serve as a regulatory checkpoint.Both DNA template sequence and polymerase structure can modulate this pause. Recent work on elongating RNA polymerase by multiple groups identified a sequence element (G-10Y-1G+1) that promotes pausing during elongation. We have found that the ‘YG’ portion of this sequence can induce a long-lived pause (∼20s) during early initiation on a lac promoter. We also identify a structural region of the polymerase, σ3.2, which contains a loop blocking the RNA-exit channel, as a pausing determinant. This region blocks RNA extension but also forms stabilizing contacts with the nascent RNA that may serve to limit transcript dissociation.The observation of a ‘YG’-induced pause during both initiation and elongation suggests that its structural basis is conserved in the initiation-elongation transition -- in agreement with the idea that the ‘YG’ motif biases the polymerase towards the pre-translocated state. Indeed, our results with a σ3.2 deletion mutant suggest that its inability to maintain the translocational register of the nascent RNA transcript prevents ‘YG’-induced pausing.We have used the information obtained on pausing to develop a new working model for initial transcription. In addition to the insight we have gained into the dynamics and structure of transient intermediates, this project is an excellent case study in how single-molecule techniques can inform the design of classical biochemical experiments – and vice versa.