With the rapidly increasing number of older people in our societies, fall detection is becoming more important: Older adults may fall at home when they are alone and they may not be found in time for them to get help. In addition, a fall itself can cause serious injuries such as lacerations, fractures and hematomas. Although many previous studies have been reported on various fall detection technologies based on wearable sensors, the inconvenience of wearing them is problematic. Vision or ambient based methods may be alternatives, but high cost and complex installation process limit applicable areas. We propose a cost-effective, ambient-based fall detection system based on a pressure sensing triboelectric nanogenerator (TENG) array. Apart from simple observation of output signal waveforms according to different actions, key technologies, including appropriate filtering and distinguishing between falls and daily activities, are demonstrated with data acquisition from 48 daily activities and 48 falls by eight participants. The proposed system achieves a classification accuracy of 95.75% in identifying actual falls. Due to its low cost, easy installation and notable accuracy, the proposed system can be immediately applied to smart homes and smart hospitals to prevent additional injuries caused by falls.