Traditional QTL analysis utilizes linkage mapping populations created to maximize polymorphism. Often, such mapping populations are not representative of the type of germplasm commonly used in breeding. Moreover, the ability to detect marker-trait associations depends on the linkage map saturation and the magnitude of the QTL effects. Discriminant analysis (DA), a multivariate statistical procedure, has been proposed as an alternative platform to QTL analysis where ideal mapping populations and apriori linkage maps are not available as in sugarcane. In this study, a previously mapped population consisting of 100 F1 progeny derived from a S. officinarum ‘La Striped' x S. spontaneum ‘SES 147B' cross was used for DA. The population was evaluated for Brix and sucrose at the early and late plant growing seasons in 2004 and 2005. Molecular marker profiles were generated for each of the 100 individuals using the AFLP, SRAP and TRAP techniques. The population was divided into three groups (low, medium and high levels of sucrose and Brix) based on 2-standard deviation. The marker data were separated into two sets; one for each parent based on the origin of the marker and DA was performed using PROC STEPDISC and DISCRIM options of SAS for each set. In both cases, DA identified markers associated with all traits and the identified markers correctly classified (> 99%) the population into three groups. Two DA-identified markers in S. officinarum were identical to QTLs and most of the DA-identified markers pointed to the same genomic regions on the linkage map. In addition, DA identified several new markers pointing to previously unidentified regions on the linkage map. These results suggest that DA could be useful as an alternative approach to QTL analysis. Our future focus is to validate DA using a random set of diverse clones.