Introduction: SARS-2-CoV infection, or COVID-19, is characterized by a heterogeneous evolution that goes from the absence of symptoms to a very serious disease, which can result in death. To date, we do not have instruments that allow us to easily predict the level of respiratory involvement in COVID-19. We hypothesized that there would be a relationship between the frequency content of patients cough sound and the severity of their disease. Methods: 36 COVID-19 patients (17 male, 52 (44 - 56) yrs.) with different degrees of severity of pneumonia were included in this study. Spontaneous cough or that obtained after induction maneuvers were recorded. Phase 1 and 2 of 129 individual cough epochs were identified, and several frequency-based features were explored. Results: Analysis shows a significant difference between groups in which pneumonia was present and absent for both maximum frequency and the variability of the frequency in phase 1 of the cough signal, as well as between the group in which pneumonia was absent and that in which pneumonia was severe (Fig. 1). Conclusions: The results presented suggest that aspects of the frequency content of the cough sound of COVID-19 patients are affected by the disease severity. Acknowledgment: This work was supported in part by Grant RTI2018-098472-B-I00 MCIU/AEI/FEDER, UE.