Sugarcane (Saccharum spp.) is a very important crop for sugar and alcohol production in Brazil. A great progress on its genome has emerged through molecular markers, which have been widely used for several purposes, such as QTL mapping based on bi-parental crosses. However, populations derived from designed crosses tend to have limited time for recombination, resulting in relatively low mapping resolution. In this sense, association mapping has emerged as a promising tool for QTL detection, by exploiting historical and evolutionary recombination events. To define appropriate strategies for association mapping, a key feature is to account linkage disequilibrium (LD) and population structure for control unlinked LD. Here we measured LD and assessed population structure in the Brazilian panel of sugarcane varieties (BPSV), composed by 135 individuals. A total of 1,474 polymorphic markers was scored, considering 86 EST-SSRs and 14 SSRs primers. A previous genetic map was used to verify distances between mapped BPSV markers. Fisher exact test was performed for all pairs of markers and used as a LD measure. Population structure was assessed by STRUCTURE and Neighbor-Joining, based on simple matching dissimilarity between all pairs of individuals. Strong LD was observed up to 15cM, mainly within the first 5cM, suggesting a higher genome coverage with markers to increase mapping resolution. Four subpopulations were detected with both methods, which is in agreement with the knowledge of germplasm by brazilian breeders. These results can provide directions for future studies of genetic mapping, which would certainly need to consider high ploidy of sugarcane.