This paper studied the detection of falls and activities of daily living (ADL) with two objectives: (1) minimum number of sensors for a broad range of activities and (2) maximize the comfort of the wearer for long term use. We used a garment to provide long term comfort for the wearer, with a 3-axis MEMS accelerometer on the shoulder position, as a wearable platform. ADL were detected in time-frequency domain and summation of absolute peak values of 3-D acceleration signals was used as feature in fall detection. 6 male and female subjects performed approximately five-hour long experiment. Sensitivity of 94.98% and specificity of 98.83% for altogether 1495 activities were achieved. Our garment-based detection system fulfilled the objective of providing the comfort of the wearer in long term monitoring of falls and ADL with high sensitivity. In fall detection, our device can summon medical assistances via SMS (Short Message Service). This detection system can raise fall alarm (fall SMS) automatically to individuals to get a shortened interval of the arrival of assistance.