Genetic engineering as an important approach to strain optimization has received wide recognition. Recent advances in the studies on the biosynthetic pathways and gene clusters of Streptomyces make stain optimization by genetic alteration possible. Kanamycin B is a key intermediate in the manufacture of the important medicines dibekacin and arbekacin, which belong to a class of antibiotics known as the aminoglycosides. Kanamycin could be prepared by carbamoylkanamycin B hydrolysis. However, carbamoylkanamycin B production in Streptomyces tenebrarius H6 is very low. Therefore, we tried to obtain high kanamycin B-producing strains that produced kanamycin B as a main component. In our work, aprD3 and aprD4 were clarified to be responsible for deoxygenation in apramycin and tobramycin biosynthesis. Based on this information, genes aprD3, aprQ (deduced apramycin biosynthetic gene), and aprD4 were disrupted to optimize the production of carbamoylkanamycin B. Compared with wild-type strain, mutant strain SPU313 (ΔaprD3, ΔaprQ, and ΔaprD4) produced carbamoylkanamycin B as a single antibiotic, whose production increased approximately fivefold. To construct a strain producing kanamycin B instead of carbamoylkanamycin B, the carbamoyl-transfer gene tacA was inactivated in strain SPU313. Mutant strain SPU314 (ΔaprD3, ΔaprQ, ΔaprD4, and ΔtacA) specifically produced kanamycin B, which was proven by LC-MS. This work demonstrated careful genetic engineering could significantly improve production and eliminate undesired products.