Poinsettia is a popular and important ornamental crop, mostly during the Christmas season. Its bract colouration ranges from pink/red to creamy/white shades, with nearly all white varieties being obtained through mutation breeding (γ- or X-ray mutagenesis) of red varieties. The appearance of acyanic poinsettia varieties is referred to as the ‘white paradox’ since the expression of most structural genes and the related enzyme activities involved in the formation of red anthocyanin pigments are present. To unravel the genetic factors responsible for the ‘white paradox’, we firstly assembled and functionally annotated a hybrid de novo bract transcriptome for the species, as well as characterized the main transcriptional differences between red- and white-bracted poinsettia varieties during bract development. The differential gene expression analysis pointed out a glutathione S-transferase (GST) gene as a putative candidate for the appearance of white genotypes. The poinsettia GST (named Bract1) is an active gene involved in the expression of anthocyanins in bracts and it presents a high phylogenetic similarity to known anthocyanin-related GSTs. We identified a 4 bp deletion in a short repeat within the coding region of Bract1, which is the most likely cause of many mutations that lead to a white bract colour. Moreover, overexpression of the Bract1 wild-type allele in Arabidopsis tt19 mutants restored the anthocyanin phenotype, while the Bract1 mutated allele showed to be non-functional. In poinsettia mutation breeding, not all red varieties can produce white sports through radiation; thus, they are distinguished into ‘heterozygous’ and ‘homozygous’ for the colouration locus according to their ability to generate white sports. The Bract1 polymorphism between wild-type and mutant alleles co-segregates with the phenotype in progeny from heterozygous red and white parents, thus confirming that Bract1 is linked to the colour trait. Based on a PCR assay for Bract1, we identified low-frequency heterozygous mutations arising from homozygous genotypes. By developing a multiplex sequencing approach, we were able to detect the presence of 1 mutated allele in a pool of 50 nonmutated copies. In conclusion, we identified a short repeat mutation in Bract1 as the most likely cause for the ‘white paradox’, which might also serve as a reference for the study of other repeat-containing structural genes as potential mutational hot spots in plant genomes. Moreover, we developed a multiplex approach that may enable an expansion of the genetic resources available for the development of new varieties.