This article deals with the recognition of early changes in the breathing pattern, in response to acute intermittent stimuli in awake rats. Two different types of stimuli were given: 9% hypoxia in N(2) and 10% hypercapnia in O(2). Animals were exposed to 3 consecutive cycles consisting of 3-min stimulus period separated by 8-min normoxic recovery intervals. Features of the breathing pattern, such as respiratory frequency, tidal volume, minute ventilation, inspiration and expiration times, peak inspiratory and expiratory flows, were measured by whole body plethysmography. The data were analyzed with the use of pattern recognition methods. We conclude that the overall respiratory changes were rather slight. However, computerized analysis using a k-nearest neighbor decision rule (k-NN) allowed for a good recognition of the respiratory responses to the stimuli. The misclassification rate (E(r)) varied from 5 to 10%. After feature selection, E(r) decreased below 1%. The k-NN classifier differentiated correctly also the type of intermittent stimulus. Our experimental results demonstrate usefulness of pattern recognition algorithms in studying respiratory effects in biological models.