The parameters influencing the respiratory pattern at or near the threshold for hearing were studied in hearing impaired and normal hearing subjects. A semiautomatic system was designed to process the data. Results indicated that thresholds for pure tones could be estimated better on the inspiration phase of the breathing cycle and that duration of the cycle was more precise than amplitude. An ascending mode of presentation was as satisfactory as the random and better than the descending methods; a 250 millisecond tone was as effective as longer tones. Estimation of thresholds were not as successful with hearing impaired as with normal hearing children; however, an accurate estimate could be made if the breathing cycle approximated a sine wave. A method based on visually apparent changes in the cycle as recorded on the strip chart was at least as accurate as the semiautomated system. A condition of visual distraction elevated auditory threshold; thresholds in the second stage of natural sleep or in sparine induced sleep were essentially similar to thresholds in the waking state. (Author/JD)