The water-soluble cell wall polysaccharides from green seaweeds of Ulva spp. (Ulvales, Chlorophyta), referred to as ulvan, demonstrate composition- and structure-related functional properties. Mid-infrared spectroscopy combined with chemometric techniques was investigated as a means to rapidly predict the chemical composition of ulvan extracts. A calibration was realized with 41 ulvan extracts from two Ulva species. The variables studied included the constituent sugars (rhamnose, xylose, glucose, galactose, glucuronic acid, iduronic acid), protein, and sulfate contents. The correlation between Fourier transform infrared and chemical data was developed using partial least squares (PLS) regression with full cross-validation (leave one out). The coefficients of determination in cross-validation ( graphic removed ) and the standard error in cross-validation were determined for each variable. The PLS model validation resulted in a coefficient of determination ( graphic removed ) and a standard error in prediction. Good predictions were obtained for rhamnose ( graphic removed = 0.9244), xylose ( graphic removed = 0.8758), glucuronic acid ( graphic removed = 0.9415), and sulfate ( graphic removed = 0.9218), which are the main ulvan constituents. However, minor components such as proteins, glucose, galactose, and iduronic acid were not correctly predicted. This study showed that mid-infrared spectroscopy combined with PLS regression is a reliable and fast method for the quantification of the main chemical constituents of ulvan extracts.