In this work, segregation data for seven commercial products at a dry mineral-based construction materials plant have been analyzed. The segregation manifests itself as an increase of the fines content to unacceptable levels at the end of complete emptying from a surge silo situated upstream of packing. Results for preliminary tests for sifting segregation and fluidization show the difficulty in estimating a priori the segregation patterns observed in practice. Knowledge of the segregation mechanisms occurring with similar materials at silo filling and the discharge flow pattern is utilized to explain the segregation at silo emptying: Accumulation of fine particles at the silo walls during filling and discharge in expanded flow cause fines-rich regions from the vicinity of walls in the lower parts of the silo to be withdrawn last. Large variations in the magnitude of segregation for each product cannot be explained for example by production run size, but are caused by size variations in the silo input, uneven distribution of fines over the heap surface during filling, unpredictable stagnant zones during discharge, and sampling procedure. Segregation at discharge mainly increases with increasing particle size distribution, but the product with the widest size distribution does not segregate the most. The width of the particle size distribution for the products correlates with the mass fraction of fine particles, and inclusion of a special additive (synthetic fibers) in very small proportions is also seen to affect segregation. Therefore, a regression model based on the mass fraction of fines and fibers was employed to organize the products according to the magnitude of segregation. Results for the model show that segregation does not always increase with an increase in the width of the particle size distribution and is negatively affected by inclusion of light and elongated particles (fibers).