The effect of changing GC content within DNA melting region as well as deletion of promoter upstream AT-rich sequences on the strength of promoter was measured in vitro and in vivo. The in vitro promoter strength was estimated by measuring the rate of formation and dissociation of promoter open complexes. At 37℃, the rate of formation of open complexes were similar for wild type, high GC, low GC, and AT-deletion mutant. At 18℃, the rate was faster in the order low GC> wild type= AT-deletion> low GC mutant. The in vivo promoter strength was estimated by measuring the activity of galactokinase (GalK) directed by promoters cloned into pK0100 plasmid in E. coli HB101. Since this plasmid contains transcriptional terminator downstream of GalK gene, we expected that the cloning of strong promoters would not affect plasmid stability significantly. At 37℃, however, all the recombinant plasmids, especially the one with low GC mutant promoter, was very unstable, and the reproducible measurement of GalK activity was not possible. At 30℃ the plasmids were relatively stable and the promoter strength was estimated to be in the order low GC> AT-deletion≥wild type> high GC mutant. Plasmid copy number was similar for each recombinant plasmids as determined by DNA hybridization and assay for β-lactamase. These results suggest that the upstream AT-rich sequences in P_R promoter does not facilitate the rate of transcriptional initiation and the GC content in the promoter melting region is another important factor in determining promoter strength.