The bacteria of the genus Streptomyces are among the most important producers of biologically active secondary metabolites. Moreover, recent genomic sequence data have shown their enormous genetic potential for new natural products, although many new biosynthetic gene clusters (BGCs) are silent. Therefore, efficient and stable genome modification techniques are needed to activate their production or to manipulate their biosynthesis towards increased production or improved properties. We have recently developed an efficient markerless genome modification system for streptomycetes based on positive blue/white selection of double crossovers using the bpsA gene from indigoidine biosynthesis, which has been successfully applied for markerless deletions of genes and BGCs. In the present study, we optimized this system for markerless insertion of large BGCs. In a pilot test experiment, we successfully inserted a part of the landomycin BGC (lanFABCDL) under the control of the ermEp* promoter in place of the actinorhodin BGC (act) of Streptomyces lividans TK24 and RedStrep 1.3. The resulting strains correctly produced UWM6 and rabelomycin in twice the yield compared to S. lividans strains with the same construct inserted using the PhiBT1 phage-based integration vector system. Moreover, the system was more stable. Subsequently, using the same strategy, we effectively inserted the entire BGC for mithramycin (MTM) in place of the calcium-dependent antibiotic BGC (cda) of S. lividans RedStrep 1.3 without antibiotic-resistant markers. The resulting strain produced similar levels of MTM when compared to the previously described S. lividans RedStrep 1.3 strain with the VWB phage-based integration plasmid pMTMF. The system was also more stable.