This paper aims to determine human blood glucose levels through analyzing the acetone present in exhaled breath as a noninvasive method with the help of an electronic nose system based on quartz crystal microbalance (QCM) sensors. The amount of acetone vapor which is the marker of blood glucose is 0.1-10 ppm in human expiration. In order for the QCM sensors to sense low levels of acetone concentration, a condenser containing zeolite absorbent ingredients is used in the experiment mechanism. The QCM sensor data obtained from breath is compared with blood glucose value. A data set of 40 volunteers with blood glucose values ranging from 84.83 mg/dl to 334 mg/dl was examined in this paper. An artificial neural network (ANN) trained using the Levenberg-Marquardt (LMNN) algorithm was developed. Data from 31 of the volunteers was used for training the ANN and data from nine volunteers was reserved for testing. Eventually, result of the study has an error of 20.13%.