Background: The Food Frequency Questionnaire (FFQ) is usually used in epidemiological studies to assess food consumption. However, the FFQ must have good accuracy, requiring its validation and reproducibility for the target population. Thus, this study aimed to describe the construction of the online Food Frequency Questionnaire (oFFQ) used at the Cohort of Universities of Minas Gerais (CUME project, Brazil) and evaluate its validity and reproducibility.Methods: The oFFQ was answered two times in 1 year (March/August 2018—March/April 2019; n = 108 participants—reproducibility), and four 24-h dietary recalls (24hRs) were applied in two seasons of the southern hemisphere [two 24hRs in autumn (March/June 2018) and two 24hRs in winter (August/September 2018); n = 146 participants—validity]. To assess the validity and reproducibility, the intraclass correlation coefficients (ICCs) were estimated.Results: The oFFQ had 144 food items separated into eight groups (dairy products; meat and fish; cereals and legumes; fruits; vegetables; fats and oils; drinks; other foods). In assessing the validity, ICCs for energy and macronutrients were considered moderate, ranging from 0.41 (energy) to 0.59 (protein), while the ICCs for micronutrients were considered low to moderate, ranging from 0.25 (fibers) to 0.65 (vitamin B6). Regarding reproducibility assessment, ICCs for energy and all the assessed items were considered moderate to excellent, ranging from 0.60 (vegetables) to 0.91 (vitamin E and retinol).Conclusions: The self-reported oFFQ had satisfactory validity and reproducibility. So, it can be used to analyze the association between food consumption and chronic diseases in the participants of the Cohort of Universities of Minas Gerais (CUME project—Brazil).