Aerodynamic measures are frequently used to analyse and document pathological voices. Some normative data are available for speakers from the English-speaking population. However, no data are available yet for Chinese speakers despite the fact that they are one of the largest populations in the world. The high variability of aerodynamic measures between and within subjects raises the issue of reliability and usefulness of this procedure in discriminating between normal and pathological voices. This study aimed to investigate the use of mean airflow rate and sub-glottal pressure in predicting normal and pathological voices. It also aimed to determine whether a higher number of repeated airflow measures would provide a better representation of the normative data in distinguishing normal from abnormal voices. The study provided a small set of preliminary normative data for Chinese speakers. Aerodynamic measures were collected from 56 Cantonese female adults using a Kay Elemetrics Aerophone II. The results showed that the accuracy in predicting a voice to be dysphonic or normal using aerodynamic measures was as high as 91.1%. The accuracy was found to improve when the number of measurements for each aerodynamic parameter was increased from three to five. The overlapping of data between the dysphonic and non-dysphonic groups, however, suggests that the aerodynamic measures should be used as an adjunct to assessment of voice disorders rather than as a diagnostic tool alone.