Most studies on the early detection of laryngeal pathology have been based upon perception. The few acoustical studies performed have been based upon pitch perturbation, amplitude perturbation, or inverse glottal wave filtering. Each of these techniques has rather severe limitations depending upon the degree of the pathology being studied. In this study a new measure is introduced for early detection of laryngeal pathology based upon a linear speech production model. The signal of interest is referred to as a residue signal and is obtained by extracting both the vocal tract characteristics (as with inverse glottal wave filtering) and the smoothed or trend characteristics of the glottal volume velocity and lip radiation characteristics. The analysis technique is easily automated and even though it does not resolve difficult problems such as source‐tract inseparability for severe cases, it has been demonstrated to be a fairly sensitive indicator of laryngeal disease in a study of ten normal and ten pathological speakers. Physical properties of this technique will be given along with examples. [This research has been supported by the National Institutes of Health under Grant No. RO‐NB‐08177.]