In current medical practice, all transmitters used for monitoring an ECG send data continuously. If transmission could be limited to only the ECG data that requires the attention of a technician, the likelihood of interference and power consumption would be reduced considerably. In this paper, we propose a new ECG telemetry system. The transmitter has a simple ECG analyzer and it only transmits data when abnormal waveforms are detected. As part of this system, we propose a simple algorithm to detect an abnormal ECG. The algorithm is composed of a baseline drift canceller (utilizing a moving average calculation), a waveform detector (using a modified second order derivative) and an ECG analyzer. In order to test the effectiveness and feasibility of the algorithm, computer simulations (with a MIT-BIH ECG database) and experiments (with microprocessor-based hardware) were performed. The results of the simulations show that the detection performance and data compression performance of the algorithm are adequate. The results of the experiment show that the algorithm even performs adequately when used with microprocessor-based hardware that is of limited size or limited power consumption.