Latent vector analysis is a rigorous mathematical technique specially suited for extracting information from cross-correlated data such as that often used to describe pulp quality. This technique has been used to construct models which relate more than 75% of the variations in hand-sheet properties to changes in both physical and chemical intrinsic fibre characteristics. Variations in over 30 properties ranging from standard handsheet tests to fibre flexibility and pentosan index, are expressed in terms of only 5 composite quality indicators. These indicators describe the effects of fibre network cohesion and intrinsic fibre strength. The dominant latent vectors are related to available bonding area, fibre swelling and bond strength, as well as average fibre length. By comprehensively describing pulp quality in terms of a small number of independent indicators, latent vector analysis can also improve the efficiency of pulp quality monitoring and control.