Heavy smoking is associated with the development of chronic obstructive pulmonary disease (COPD). However, there is no valuable biomarker for evaluating COPD development in heavy smokers because they are usually asymptomatic. This study is aimed at evaluating whether the levels of serum miRNAs can serve as biomarkers for predicting the occurrence of COPD. A rat model of emphysema was induced by enforced smoking, and the dynamic miRNAs expression profile at different stages of emphysema with varying periods of smoking were analyzed by microarray and quantitative real-time polymerase chain reaction (qRT-PCR). The differentially expressing miRNAs were analyzed using Gene Ontology and the KEGG PATHWAY database. The levels of three serum candidate miRNAs were measured by qRT-PCR in 41 healthy controls (HC), 40 asymptomatic heavy smokers, and 49 COPD patients. Following smoking for varying periods, different severities of lung emphysema were observed in different groups of rats, accompanied by altered levels of some serum miRNAs associated with regulating some pathways. Furthermore, the levels of miR-21 were significantly higher in the COPD patients and asymptomatic heavy smokers than in the HC (P < 0.001), while the levels of miR-181a were significantly lower in the COPD patients and asymptomatic heavy smokers than in the HC (P < 0.001). Accordingly, the levels of serum miR-21 and miR-181a as well as their ratios had a high sensitivity (0.854) and specificity (0.850) for evaluating the development of COPD. Our data suggest that the levels of serum miR-21 and miR-181a may be valuable for evaluating the development of COPD in heavy smokers.