Styrene monooxygenases (SMOs) are two-enzyme systems that catalyze the enantioselective epoxidation of styrene to (S)-styrene oxide. The FADH₂ co-substrate of the epoxidase component (StyA) is supplied by an NADH-dependent flavin reductase (StyB). The genome of Rhodococcus opacus 1CP encodes two SMO systems. One system, which we define as E1-type, displays homology to the SMO from Pseudomonas taiwanensis VLB120. The other system, originally reported as a fused system (RoStyA2B), is defined as E2-type. Here we found that E1-type RoStyB is inhibited by FMN, while RoStyA2B is known to be active with FMN. To rationalize the observed specificity of RoStyB for FAD, we generated an artificial reductase, designated as RoStyBart, in which the first 22 amino acid residues of RoStyB were joined to the reductase part of RoStyA2B, while the oxygenase part (A2) was removed. RoStyBart mainly purified as apo-protein and mimicked RoStyB in being inhibited by FMN. Pre-incubation with FAD yielded a turnover number at 30 °C of 133.9 ± 3.5 s^{− 1}, one of the highest rates observed for StyB reductases. RoStyBart holo-enzyme switches to a ping-pong mechanism and fluorescence analysis indicated for unproductive binding of FMN to the second (co-substrate) binding site. In summary, it is shown for the first time that optimization of the N-termini of StyB reductases allows the evolution of their activity and specificity.