The calibration curve was constructed in the range of 1.05.0 mgL1 from a stock solution of HMF (1000mgL1). This narrowconcentration range was used to avoid loss of method linearity and problems in the column. Food samples analyzed were constituted of sugars and this could obstruct the column and compromise its efficiency. Three standard solutions were prepared for each point of the calibration curve and these were injected in triplicate. The linearity of the chromatographic method was checked by applying a linear regression analysis and a lack-of-fit test (Araujo, 2009; Barros Neto, Bruns, & Scarmínio, 2006; Danzer & Currie, 1998) to data of the calibration experiment at 95% confidence level. The significance of the calibration curve equation coeff icients (intercept and sensitivity of the method) was tested by a t-test. Confidence and prediction intervals were built at the same level of confidence. All statistical analyses were carried out using the statistical software Minitab 16.2.2.