**BACKGROUND: **Acute Respiratory Distress Syndrome (ARDS) is an important clinical problem. The definition of ARDS requires an arterial blood gas to define the ratio of partial pressure of arterial oxygen to fraction of inspired oxygen (PaO2/FIO2 ratio). However, many patients with ARDS do not have a blood gas measured, which may result in under-diagnosis of the condition. As a consequence, a method for estimating PaO2 from noninvasive measurements is desirable.

**METHODS: **Using data from three ARDS Network studies, we analyzed the enrollment arterial blood gas to compare non-linear to linear and log-linear imputation methods of estimating PaO2 from hemoglobin percent saturation with oxygen measured by a pulse oximeter (SpO2). We compared mortality on the basis of various measured and imputed PaO2/FIO2 ratio cutoffs to assure clinical equivalence.

**RESULTS: **We studied 1184 patients, of whom 707 had SpO2≤96%. Non-linear imputation from an SpO2/FIO2 ratio resulted in lower error than linear or log-linear imputation (p<0.001) for patients with SpO2≤96% but was equivalent to log-linear imputation in all patients. 90-day hospital mortality was 26-30%, depending on the PaO2/FIO2 ratio, whether non-linearly imputed or measured. On multivariate regression, the association between imputed and measured PaO2 varied by use of vasopressors and SpO2.

**CONCLUSIONS: **A non-linear equation more accurately imputes PaO2/FIO2 from SpO2/FIO2 than linear or log-linear equations, with similar observed hospital mortality of SpO2/FIO2 ratio versus measured PaO2/FIO2 ratios. While further refinement through prospective validation is indicated, a non-linear imputation appears superior to prior approaches to imputation.