Zymomonas mobilis is an ethanologenic bacterium that can produce hopanoids using farnesyl pyrophosphate (FPP), which can be used as the precursor by β-farnesene synthase for β-farnesene production. To explore the possibility and bottlenecks of developing Z. mobilis for β-farnesene production, five heterologous β-farnesene synthases were selected and screened, and AaBFS from Artemisia annua had the highest β-farnesene titer. Recombinant strains with AaBFS driven by the strong constitutive promoter Pgap (Pgap–AaBFS) doubled its β-farnesene production to 25.73 ± 0.31 mg/L compared to the recombinant strain with AaBFS driven by Ptet (Ptet–AaBFS), which can be further improved by overexpressing the Pgap–AaBFS construct using the strategies of multiple plasmids (41.00 ± 0.40 mg/L) or genomic multi-locus integration (48.33 ± 3.40 mg/L). The effect of cofactor NADPH balancing on β-farnesene production was also investigated, which can be improved only in zwf-overexpressing strains but not in ppnK-overexpressing strains, indicating that cofactor balancing is important and sophisticated. Furthermore, the β-farnesene titer was improved to 73.30 ± 0.71 mg/L by overexpressing dxs, ispG, and ispH. Finally, the β-farnesene production was increased to 159.70 ± 7.21 mg/L by fermentation optimization, including the C/N ratio, flask working volume, and medium/dodecane ratio, which was nearly 13-fold improved from the parental strain. This work thus not only generated a recombinant β-farnesene production Z. mobilis strain but also unraveled the bottlenecks to engineer Z. mobilis for farnesene production, which will help guide the future rational design and construction of cell factories for terpenoid production in non-model industrial microorganisms.