This paper presents two wearable electrocardiogram signal acquisition and classification systems that are capable of long-term healthcare and bio-signal record. In the first version, an MSP430 micro-control unit (MCU) combines with low-power analog front end (AFE), and a Bluetooth module is implemented. The proposed AFE consists of a low-noise pre-amplifier with gain control and a high-pass sigma-delta modulator (HPSDM) that allows the system to achieve high resolution. The direct transmission of the HPSDM outcome can enhance data security. In the other version, the MCU is replaced by a bio-signal processor (BSP). The BSP combines a decimation filter with a wavelet-transform processor to fulfill beat detection and classification. The UART package generator is also integrated in BSP. The algorithm of BSP is verified by MIT/BIH Arrhythmia data, with sensitivity and positive predictability of 99.67% and 99.59%, respectively. The entire system can be operated for 2 days by using two PR44 zinc-air batteries of 605 mAh as power supply.