Bicycle Level of Service (BLOS) provides an essential tool for evaluating the operations of low-carbon bicycle facilities and prioritizing investment in new bicycle facilities under various constraints. This study aimed at developing a LOS method for assessing bicycle facilities in the metropolitan areas of China. Using this method, we addressed major challenges in obtaining user ratings of bicycle facilities and captured senses of satisfaction of bike users riding on bicycle facilities. Virtual Reality (VR) technique was introduced to obtain data by creating 120 immersive settings or scenarios for participants. A hundred of bicyclists or participants with a wide range of characters were recruited. These participants were asked to express their senses of satisfaction under predefined physical conditions of bike facilities and traffic conditions. Their Satisfaction Rating Scores (SRS) were documented. The statistical relationships between rider’s feelings and bike facilities/traffic conditions were modeled and verified through a symbolic regression (or an effective deep learning) approach. The model is demonstrated to be reliable in predicting SRS of bicyclists with a high correlation coefficient. This study also developed a set of LOS criteria based on the cumulative distribution of satisfaction scores. These LOS criteria are simple to use and effective in assessing operational performance of existing bicycle facilities and providing decision makers with insightful guidance for planning, designing, and operating new active transportation facilities.