In raw sugar manufacture, the quality of the sugarcane supply to the factory plays the most important role in production costs. Sugarcane can be very susceptible to damage by freezes, and freeze-deteriorated sugarcane can cause problems in processing which sometimes leads to a factory shut-down. A reliable indicator of whether a certain shipment of sugarcane can be processed economically is still needed. This study was undertaken during the 2002/2003 harvest season to measure the cold-tolerance performance of eight commercial sugarcane varieties, and to establish indicators which can predict processing quality and problems. Varieties studied included CP 70-321, CP 79-318, LHo 83-153, LCP 85-384, HoCP 85-845, HoCP 91-555, HoCP 96-540 a newly released Louisiana variety, and TucCP 77-42, developed in Argentina. Freezing temperatures occurred between January 18–19 and 24–25, 2003. The minimum field temperature recorded was −5.1 °C on January 25. Freezing conditions prevailed for 10–14 h during each freeze. Serious tissue damage occurred within the stalks of all varieties following the January 24–25 freeze, including freeze cracks. Samples were taken one day before the first freeze (pre-freeze control) on January 17, and subsequently again 12, 18 and 26 days post-freeze. Marked changes for most indicators of freeze-deterioration, for all varieties, were observed, particularly 18 and 26 days post-freeze, and viscosity increased and percentage pol-filterability decreased on freeze-deterioration. Variety TucCP 77-42 had significantly ( P<0.05) the worst cold-tolerance, even after 12 days. Strong polynomial trends or quadratic fits existed between ASI-II dextran and titratable acidity ( r 2=0.916) and pH ( r 2=−0.883). Deterioration effects became greater than varietal effects at threshold levels of ∼2500 and ∼2800 ppm/Brix dextran for titratable acidity and pH, respectively. Titratable acidity and pH may, therefore, only be useful in predicting problems caused by severe dextran concentrations. Reactions of Leuconostoc mesenteroides of importance to sugarcane deterioration, including the production of dextran, levan, and alternan polysaccharides are fully described. Mannitol, produced by mannitol dehydrogenase from Leuconostoc, was a slightly better predictor of viscosity ( r 2=0.838) than both ASI-II ( r 2=0.802) and Haze ( r 2=0.814) dextran, because it can indicate all Leuconostoc polysaccharides. In comparison, ethanol ( r 2=0.676), leucrose ( r 2=0.711), and pH ( r 2=−0.711) were only moderately correlated with viscosity. Mannitol was also better than dextran at predicting percentage pol-filterability, although substances present in the undeteriorated cane juice and sucrose interfere with this processing parameter. Overall, mannitol was the best predictor of sugarcane deterioration which contributes to sucrose losses, dextran, viscosity, and to a lesser extent, filterability problems. Model mannitol degradation reactions, simulating industrial conditions, showed that no mannitol degradation occurred, even after 1 h at 99 °C, pH 5.4 and high or low Brix, which further supports the use of mannitol to indirectly measure dextran and/or deterioration in the factory.