The COVID-19 is a disease that presents a wide variety of combinations and intensities of symptoms, characteristic of a Flu Syndrome (FS), which can quickly evolve to a Severe Acute Respiratory Syndrome (SARS). The objectives of this study were to evaluate the hierarchy of symptoms of FS in patients with SARS caused by COVID-19 and to develop a prediction model for potential cases based on sex and race. Binary logistic regression modeling was used in 405,419 records selected from the database of the Ministry of Health of Brazil. It was found that men were more affected by the disease, with a 15.5% higher risk than women. They also died more, with a 13.8% and 15% higher risk for all causes and for COVID-19, respectively. The chances of more than one non-white patient dying from all causes ranged from 18.4% to 38.7% and for Covid-19 it ranged from 16.7% to 64.3% according to race. Fever, muscle pain and loss of smell or taste alternate in the first three positions of the symptom hierarchy, according to sex and race. Cough was only relevant for white men and sore throat for black men. Vomiting was only relevant for black women. The best prediction model developed encompassed seven symptoms adjusted for age, sex and race, but was able to explain only 63% of the cases of COVID-19. Possibly racial diversity, and the socioeconomic inequality associated with it, make the challenge of estimating probabilities of infection by COVID-19, based on symptoms, more complicated in Brazil than in other countries.