Aim To investigate the serum expressions of miR-663, miR-126 and miR-370 of microRNA (miRNA) after stent implantation in patients with coronary heart disease, and to analyze the predictive value of the above 3 types of miRNA in in-stent restenosis (ISR). Methods The study included 175 patients with coronary heart disease, who underwent coronary stent implantation and followed up coronary angiography, and the patients were divided into ISR group (n＝66) and non-ISR group (n＝109) according to the results of angiography. The clinical data were analyzed and compared in the two groups. The serum samples of all patients were collected and the expression levels of serum miR-663, miR-126 and miR-370 were detected by real-time quantitative PCR. The values of the above 3 types of miRNA predicting ISR were analyzed by statistical methods. Results A total of 66 cases of ISR occurred in 175 patients with coronary heart disease after stent implantation. Smoking rate was higher in ISR group (P＜0.05). The levels of blood glucose, low density lipoprotein cholesterol and high-sensitivity C-reactive protein (hs-CRP) in ISR group were higher than those in non-ISR group (P＜0.01). The expressions of serum miR-663 and miR-126 in ISR group were lower than those in non-ISR group, but the expression of miR-370 was higher than that in non-ISR group (all P＜0.01). Logistic regression analysis showed that the expression levels of miR-663 and miR-126 were related to ISR, and smoking and hs-CRP were independent risk factors of ISR. The results of receiver operating characteristic curve showed that the area under curve (AUC) of miR-663, miR-126 and miR-370 predicting ISR were 0.2,0.720 and 0.650, respectively, while the AUC of miR-663 combined with miR-126 predicting ISR was 0.909, which was superior to the prediction efficiency of single miRNA. Conclusion The expressions of miR-663 and miR-126 were decreased and the expression of miR-370 is increased in patients with ISR after coronary stent implantation. Serum miR-663, miR-126 and miR-370 can be used to predict the occurrence of ISR, and miR-663 combined with miR-126 can significantly improve the efficacy of ISR prediction.