The low signal-to-noise ratio encountered during auscultation in many high noise environments can impede a physician’s successful examination and diagnosis of a patient’s health. This thesis develops a vibro-acoustic model for an electronic stethoscope and investigates a number of techniques to improve the signal-to-noise ratio. The techniques explored are: 1. Redesign of stethoscope components for improved vibration isolation 2. Use of dual piezoelectric transducers and dynamic model inversion for elimination of physician handling noise 3. Implementation of active noise cancellation using either a reference microphone or a reference accelerometer In a digital stethoscope, a piezoelectric transducer is used to convert chest sounds into electrical signals. Due to the larger chestpiece size needed to accommodate the electronics of a digital stethoscope, noise due to physician handling of the device is often greater. To characterize the effects of the device’s construction on its sensitivity to handling disturbances, a theoretical stethoscope model is developed. The vibro-acoustic model relates force inputs acting on the body of the stethoscope to voltage signals created by its piezo-ceramic transducer. Using the theoretical model, simulations are conducted to demonstrate that traditional vibration isolation applied between the chestpiece and the transducer results in poor coupling—loss in sensitivity—between the transducer and the patient’s chest. By using a floating transducer housing, equally effective vibration isolation can be accomplished with far less loss in sensitivity to patient signals. The proposed isolation design was experimentally evaluated by redesigning the stethoscope’s components. However, limitations in the damping and stiffness values of available isolation materials resulted in some loss of sensitivity over a narrow frequency range. Next, as a superior alternative to physically isolating the stethoscope transducer from external disturbances a signal processing based approach to compensate for handling noise is developed. It is possible to use a redundant sensor and novel input estimation techniques to digitally remove the undesired noise measurement components. By adding a second piezo to the stethoscope assembly, it is shown that an inverse dynamic mapping can be used to relate the measured signals to original directional inputs acting on the stethoscope. An output feedback observer is developed to account for the unknown initial state of the system dynamics. In simulation, it is shown that the effects of the unknown and undesired disturbance input can be removed over the entire frequency range critical for auscultation. In physical experiments, the feasibility of the dual-piezo stethoscope approach to estimate and remove these disturbances is also demonstrated. In many patient transport environments, the ambient noise can routinely exceed 75 dB with the most severe environments having noise sources more than 3000 times louder than a typical auscultation signal. For noise from the patient side of the stethoscope, passive isolation methods cannot be used as they will impede transmission of the desired chest sounds to the transducer. Based on helicopter field data acquired with an electronic stethoscope retrofitted with an array of microphones and accelerometers, it is demonstrated that primary noise corruption during auscultation in a helicopter can be attributed to vehicle vibrations. Using this information, it is shown that a reference accelerometer can be used in place of a conventional reference microphone to estimate the noise corruption. Using either the LMS or NLMS active noise cancellation algorithms, it is possible to extract the desired auscultation signal. This is confirmed experimentally by simulating the internal cabin noise levels of the harshest noise environment in which injured military personnel are routinely examined—a mobile Black Hawk helicopter.