Tuber shape is an intriguing morphological trait, which displays continuous trait variation ranging from flat, round to oval and long. Initially a single locus model was proposed to explain tuber shape, where multiple alleles rather than multiple loci were proposed to explain quantitative variation (van Eck et al. 1994). Besides this major-effect QTL on chromosome 10 another minor effect QTL has been published on chromosome 2, explaining 8% of the variance (Prashar et al. 2014). To obtain a better overview on the loci contributing to variation in tuber shape a comprehensive genome wide association study (GWAS) was performed in a panel of 537 commercial potato cultivars. This confirmed that the Ro locus is the major-effect QTL, but also the minor effect QTL on chromosome 2 was found. In addition, on chromosome 10, colocalization of the major effect QTL for tuber shape and a major QTL for eye depth was observed. For the Ro locus, most significant associations were found to localize on superscaffold PGSC0003DMB000000385 on chromosome 10. To refine this region we performed a recombinant screening in a diploid population (C×E). Recombinant analysis resulted in the identification of 104 recombinants originating from the female meiosis and 27 recombinants from the male meiosis. Recombinant analysis with additional SNPs within the selected region allowed us to confine the Ro locus to a 280 kb region, located on superscaffold DMB546 (323kb). Within this region a cluster of cell wall III peroxidase genes is found. Based on the putative role of peroxidases, this gene family cluster of repeats is likely to be implicated in mediating differences in tuber shape. Keywords Solanum tuberosum, organ morphology, high resolution mapping, GWAS, haplotypes Conclusion: We present a scalable approach that accurately reconstructs haplotypes in polyploid crops. The resulting haplotypes are instrumental for analysing the haplotype composition of the potato gene pool, and haplotype based QTL discovery.