Chardonnay dry white wines from main wine producing districts in China were collected to analyze aroma quality and reveal the mathematic relations between aroma characters and volatiles. Sensory analysis was made by 30 panelists who were trained by using “Le Nez du Vin” wine aroma kit. The sensory data were composed of intensity and by using frequency of characters. SPME—GC—MS was used to identify and quantify aroma compounds in wine. PCA and correlation analysis of sensory data showed six typical aroma characters of the Chinese Chardonnay dry white wines: citric, pineapple, violet, lime, green apple and lemon. Odor active value (OAV) analysis indicated that 25 ones of 64 aroma compounds quantified in sample wines were impact odorants. Six regressive models of typical characters were built by aroma compounds to predict typical aroma intensity. Aroma quality and wine origin could also be estimated by the models. Five models have significance at 0.01 levels on ANOAV. Coefficients and constants of models are also significant at 0.01 level, and the correlation coefficients between the measured and predicted values （R2） are between 0.979 and 0.999. Chinese Chardonnay dry white wines from main producing districts have typical aroma characters. Aroma characters and volatile compounds have obvious relative relations, and some aroma compounds could be used to predict wine aroma characters.