Recently, the detection of occult cancer cells in peripheral blood has received a great deal of attention regarding the prediction of postoperative cancer recurrence and for novel strategies of adjuvant therapy. The aim of this study was to establish a new molecular diagnostic method of detecting circulating tumor cells. Gastric cancer SGC-7901 cells in 2 ml blood from healthy volunteers were serially diluted. Additional peripheral blood samples were collected from 90 patients and 27 healthy volunteers. Real-time reverse transcription-polymerase chain reaction was used to detect the levels of microRNA-106a (miR-106a) and microRNA-17 (miR-17). Receiver operating characteristics (ROC) curves were constructed. In recovery experiments, a significant correlation between the number of cancer cells and the levels of both miR-106a (r = -0.906, p = 0.037) and miR-17 (r = -0.912, p = 0.031) was found. In preoperative and postoperative patient groups, miR-106a and miR-17 levels were significantly higher than those in controls. The areas under the ROC curve for miR-106a, miR-17, and combination were 0.684 (p = 0.0066), 0.743 (p = 0.0001), and 0.741 (p = 0.0002), respectively. Our results indicate that the detection of miRNA in peripheral blood may be a novel tool for monitoring circulating tumor cells in patients with gastric cancers.