The present invention relates to a method and a system designed to determine an individual's state of attention from an individual's respiratory signal. The method of the invention comprises a first learning step, wherein the characteristics of an individual's normal state are determined by selecting a respiratory signal fragment that is considered to be normal according to a pre-established criterion, and a second analysis phase, wherein the individual's state of attention is determined from parameters extracted from the respiratory signal on the basis of some pre-defined rules and the individual's normal state previously characterised. The method of the invention is implemented in two embodiments. A first embodiment is based on the identification of pre-defined patterns in the respiratory signal and, in the second embodiment, an index indicative of the variability of the respiratory signal is defined.