This research investigated the use of acoustic and electroglottographic (EGG) analysis in support of the diagnosis of pathologies affecting the larynx and voice production. Experiments were conducted (1) to verify the reliability of perturbation measures from sustained vowels and (2) to obtain normative perturbation values in a dysphonic population. The studied measures included jitter, shimmer, signal-to-noise ratio, EGG closed-quotient, and EGG speed index. Although such parameters have been widely studied in the field of voice production, their reliability has been investigated mostly on non-dysphonic speakers, synthetic signals, or artificially produced perturbations. Synthetic stimuli may permit well-controlled systematic studies, but still fail in adequately reproducing the vocal perturbations found in dysphonic speakers. Therefore, many reported normative values and computational algorithms need to be validated on real disordered voices before being recommended for clinical use. Some computational algorithms are being proposed in this thesis, including a robust and reliable method for measurement of acoustic jitter, a procedure for estimating the signal-to-noise ratio on a cycle-to-cycle basis, and a highly accurate EGG fundamental frequency tracker. Practical protocols for data acquisition in a clinical environment are also being proposed. The most important contribution of this research was the development of a method for detecting certain asymmetries in the posterior part of the larynx. These problems were commonly found in the "no-abnormality-detected" and "functional" diagnostic groups. The detection method exploits a differential increase in EGG jitter values that was induced, possibly, by instabilities in the positioning of the arytenoid cartilages during phonation. These asymmetries, which can be caused by unilateral paralysis of the superior laryngeal nerve, among other causes, seemed particularly harmful to singers and other professional voice users.