Finding the best chip size distribution can help reduce strength loss during the winter. The objective was to assess the impact of chip length and thickness on the properties of handsheets made from ultra-high-yield sulfite pulp, by applying multivariate statistical analysis to data archived over 16 months. An empirical model linking chip size fractions to handsheet quality was constructed. This served as the basis for a numerical optimization in which the chip size distribution that simultaneously maximizes tensile energy absorption (TEA), tear index and burst strength was identified. Background information leads on to a description of the multivariate statistical modeling process. A partial least squares (PLS) model was constructed from mill data. Four conclusions are summarised, including the fact that almost 40% of pulp quality variability relates to seasonal changes in chip feed characteristics. The effects of the addition of fine particles and lower chip densities are explained.