Two samples of contaminated coffee (5.3 and 57.2 ppb of OTA) were roasted at 230°C using two methods: rotating cylinder (RC) and fluidized bed (FB). Samples were taken every 3 and 0.9 min from RC and FB respectively. Each sample was analyzed for OTA content. The results showed that the process by RC was more effective on OTA reduction than FB for a same degree of roasting with 67% OTA reduction against 36% for the medium roast. The thermal degradation rate of pure OTA and the OTA mixed with the components of coffee (5 sugars, 3 amino acids, caffeine and chlorogenic acids) were determined, showing that interactions took place dependent themselves on the conditions of pH and pKa values of the components tested, in this case by influencing by the reactivity and the rate of degradation of OTA. A transformation product (TP) was observed in the chromatograms obtained from the interaction of OTA with the components of coffee. A test of alkalinization and warming of pure OTA confirmed that the TP comes from the structural modification of the OTA molecule and is not a product of interaction with the natural components of coffee. The TP was purified to carry out its chemical elucidation. The chemical nature of compound transformation and spectroscopic data such as UV-Vis (?max: 237nm), the affinity with the mobile phase of the OTA, the analysis of alkalinization (OTA regeneration phenomenon and TP), the analysis of stable isotopes (SIDA's) and the mass spectrum (molecular ion M +: 420 m/z), suggest that the TP of OTA during the roasting process corresponds structurally to an analogue of OTA which retains its acidic carboxyl group and in accordance to fragmentation corresponds to the Hydroxi-Ochratoxin A (OH-OTA), as well as minor amounts of OTA and its isomers.