Among cancer patients treated with fluoropyrimidines, 10–40% develop severe toxicity. Polymorphism of the dihydropyrimidine dehydrogenase (DPYD) gene may reduce DPD function, the main enzyme responsible for the metabolism of fluoropyrimidines. This leads to drug accumulation and to an increased risk of toxicity. Routine genotyping of this gene, which usually includes DPYD *HapB3, *2A, *13 and c.2846A > T (D949V) variants, helps predict approximately 20–30% of toxicity cases. For DPD intermediate (IM) or poor (PM) metabolizers, a dose adjustment or drug switch is warranted to avoid toxicity, respectively. Societies such as the Spanish Society of Pharmacogenetics and Pharmacogenomics (SEFF), the Dutch Pharmacogenetics Working Group (DPWG) or the Clinical Pharmacogenetics Implementation Consortium (CPIC) and regulatory agencies (e.g., the Spanish Medicines Agency, AEMPS) already recommend DPYD routine genotyping. However, the predictive capacity of genotyping is currently still limited. This can be explained by the presence of unknown polymorphisms affecting the function of the enzyme. In this case-control work, 11 cases of severe fluoropyrimidine toxicity in patients who did not carry any of the four variants mentioned above were matched with 22 controls, who did not develop toxicity and did not carry any variant. The DPYD exome was sequenced (Sanger) in search of potentially pathogenic mutations. DPYD rs367619008 (c.187 A > G, p.Lys63Glu), rs200643089 (c.2324 T > G, p.Leu775Trp) and rs76387818 (c.1084G > A, p.Val362Ile) increased the percentage of explained toxicities to 38–48%. Moreover, there was an intronic variant considered potentially pathogenic: rs944174134 (c.322-63G > A). Further studies are needed to confirm its clinical relevance. The remaining variants were considered non-pathogenic.