The early detection of laryngeal diseases increases the survival rate of laryngeal cancer and prevents radical destructive surgery. Also, it is essential to cure laryngeal cancer. The diagnosis of laryngeal diseases using ARS has many advantages being simple, fast, non-invasive. It requires a recording sound equipment in a samll space. In this study, we strived to obtain parameters to help diagnose laryngeal diseases accurately. we evaluated the voice data collected by ARS from 119 laryngeal disease patients as well as normal control group who visited the otolaryngology department of PNUH from December, 1999 to June, 2000. Twelve acoustic parameters were determined after observing the distribution and analyzing them statistically. Hit ratio was obtained by artificial neural network. Eight acoustic parameters were found to be essential differentiate to normal from laryngeal diseases group. We found that ARS can be used to diagnose laryngeal diseases and the parameters and detection programs lead to robust output.