Coronaviruses use discontinuous transcription to generate subgenomic RNAs (sgRNAs) that encode structural and accessory proteins. However, the factors regulating sgRNA abundance in SARS-CoV-2 remain unclear. Here, we combined strand-specific RNA sequencing, RNA–RNA interaction mapping, prediction of RNA folding energies, and targeted mutagenesis to define the regulation of (–) sgRNA synthesis in SARS-CoV-2 infection. We demonstrated that the relative (–) sgRNA abundance across viral genes is stable throughout infection and largely correlates with corresponding (+) sgmRNA levels. Through meta-analysis of published SPLASH data, we found that the frequency of long-range interactions between the 5′ genomic transcription regulatory sequence TRS-Leader and downstream TRS-Body sequences correlates with sgRNA abundance.