Large polyploid genomes of non-model species remain challenging targets for DNA polymorphism discovery despite the increasing throughput and continued reductions in cost of sequencing with new technologies. For these species especially, there remains a requirement to enrich genomic DNA to discover polymorphisms in regions of interest because of large genome size and to provide the sequence depth to enable estimation of copy number. Various methods of enriching DNA have been utilised, but some recent methods enable the efficient sampling of large regions (e.g. the exome). We have utilised one of these methods, solution-based hybridization (Agilent SureSelect), to capture regions of the genome of two sugarcane genotypes (one Saccharum officinarum and one Saccharum hybrid) based mainly on gene sequences from the close relative Sorghum bicolor. The capture probes span approximately 5.8 megabases (Mb). The enrichment over whole-genome shotgun sequencing was 10–11-fold for the two genotypes tested. This level of enrichment has important consequences for detecting single nucleotide polymorphisms (SNPs) from a single lane of Illumina (Genome Analyzer) sequence reads. The detection of polymorphisms was enabled by the depth of sequence at or near probe sites and enabled the detection of 270 000–280 000 SNPs within each genotype from a single lane of sequence using stringent detection parameters. The SNPs were present in 13 000–16 000 targeted genes, which would enable mapping of a large number of these chosen genes. SNP validation from 454 sequencing and between-genotype confirmations gave an 87%–91% validation rate.