A new smart fall-down detector for senior healthcare system using inertial microsensors and Wi-Fi technology has been designed, prototyped and characterized in this work. The detector can reduce the risk of severe injury or death caused by falling down with minimum false alarm rate. The different patterns of motion are sensed by a set of inertial sensors composed of a tri-axial accelerometer and a tri-axial gyroscope. The signals of motion are sampled and processed by a microcontroller with integrated algorithms. The smart algorithm integrated with machine learning can be customized according to different habits of different seniors to reduce false alarms. The fall-down signal is transmitted through Wi-Fi to the client via Internet.