Aim: To investigate whether exhaled carbon monoxide (CO) is dose-dependently associated with subgingival polymicrobial clusters before and after non-surgical periodontal therapy. Material and Methods: We followed 163 adults with periodontitis, measuring CO before and at 2, 14 and 26 months after therapy. Subgingival plaque samples were analysed using 16S rRNA gene sequencing. We applied topic models and regression analysis, adjusted for age, sex, treatment group, pocket probing depth and plaque levels. Results: Smokers had generally higher mean CO levels at every visit and showed significant declines at 14 and 26 months after therapy (p < 0.01). Before therapy, CO was not associated with dysbiosis; after therapy, higher levels were positively associated. In sensitivity analyses, self-reported smoking status explained more variance in dysbiosis than CO alone, with only marginal gains when both were combined. Among smokers, a distinct polymicrobial cluster dominated by Fusobacterium and Prevotella increased with higher exhaled CO. Every 10 parts per million increase in CO was associated with an estimated 16% increase in cluster abundance (95% confidence interval: 8%–26%). Conclusion: Exhaled CO may help monitor persistent dysbiosis and select an anaerobe-dominated subgingival community after periodontal therapy among smokers.