Introduction Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) outbreaks have led to massive admissions to intensive care units (ICUs). An ultrasound examination of the thorax is widely performed on admission in these patients. The primary objective of our study was to assess the performance of the lung ultrasound score (LUS) on ICU admission to predict the 28-day mortality rate in patients with SARS-CoV-2. The secondary objective was to asses the performance of thoracic ultrasound and biological markers of cardiac injury to predict mortality. Methods This multicentre, retrospective, observational study was conducted in six ICUs of four university hospitals in France from 15 March to 3 May 2020. Patients admitted to ICUs because of SARS-CoV-2-related acute respiratory failure and those who received an LUS examination at admission were included. The area under the receiver-operating characteristics (ROC) curve was determined for the LUS score to predict the 28-day mortality rate. The same analysis was performed for the Simplified Acute Physiology Score, left ventricular ejection fraction, cardiac output, brain natriuretic peptide and ultra-sensitive troponin levels at admission. Results In 57 patients, the 28-day mortality rate was 21%. The area under the ROC curve of the LUS score value on ICU admission was 0.68 [95% CI 0.54-0.82; p = 0.05]. In non-intubated patients on ICU admission (n = 40), the area under the ROC curves was 0.84 [95% CI 0.70-0.97; p = 0.005]. The best cut-off of 22 corresponded to 85% specificity and 83% sensitivity. Conclusions LUS scores on ICU admission for SARS-CoV-2 did not efficiently predict the 28-day mortality rate. Performance was better for non-intubated patients at admission. Performance of biological cardiac markers may be equivalent to the LUS score.