The acute respiratory distress syndrome (ARDS) consists on a lack of breath due to fluid overload in the lungs that is not produced by a heart desease. Some people with this condition may need to be intubated and connected to invasive mechanical ventilation, but less severe cases may need supplementary oxygen that can be delivered with non-invasive devices, such as CPAP (continuous positive airway pressure) or HFNO (high flow nasal oxygenation). CPAP consists on a facemask that provides oxygen-enriched air at a high pressure, whereas HFNO consists on nasal cannula that provides oxygen-enriched air at a high flow. Patients with ARDS may present with high respiratory efforts that can eventually damage their own lungs and contribute to the development of a phenomenon known as patient self-inflicted lung injury (P-SILI). Previous research has identified that CPAP may be successful in attuenuating P-SILI compared to HFNO, but it is not known whether this attenuation actually results into a reduction in lung injury in real patients. In this multicentre trial, 120 non-intubated patients with stablished ARDS will be randomly assigned to receive oxygen-enriched air through either CPAP or HFNO for 48 hours plus standard intensive care. The primary goal of this study is to determine the pulmonary effect of CPAP and HFNO through lung injury biomarkers that can be detected in blood, such as sRAGE (soluble Receptor of Advanced Glycation End-products), angiotensin-II, interleukin-6 and interleukin-10. It will also be studied whether CPAP reduces 48-hour traqueal intubation rate, 90-day traqueal intubation rate and 90-day mortality. Identifying that CPAP attenuates lung injury in spontaneously breathing ARDS patients will help clinicians to better understand this condition and to better treat this patients, so they do not evenutally need traqueal intubation and connection to invasive mechanical ventilation.