Abstract Background Right atrial pressure (RAP) is a key metric in haemodynamic assessment. Elevated RAP is associated with poor prognosis in heart failure and pulmonary hypertension and physiologically with fluid overload. RAP can be measured invasively or non-invasively, but Cardiac MRI (CMR) currently cannot estimate RAP. Purpose To develop a model to estimate RAP from CMR from paired right heart catheter (RHC) and CMR assessments. Assess correlation to WHO functional status prior to prospective clinical validation studies. Methods Patients were recruited to a registry of those referred for assessment of dyspnoea between 2012 and 2020. Inclusion criteria were age >18 years, signs and symptoms of heart failure and adequate CMR image quality. Patients diagnosed with pulmonary arterial hypertension were excluded. RHC and CMR were performed on the same day. CMR 2- and 4- chamber cines were used to measure chamber dimensions, strain, ejection fraction and stroke volume with operator reviewed AI contours in MASS research software. Pearson’s product-moment correlation coefficient (r ) was used to assess relationships between CMR metrics and invasive mRAP. Variables were then included in stepwise multiple linear regression models to predict mRAP. These were compared with receiver-operator curve analysis and DeLong’s test. Kruskal-Wallis test compared median mRAP between WHO functional class groups. Results The cohort of 672 patients was divided based upon invasive mRAP ≤ 8mmHg (44%) and mRAP > 8mmHg (56%). Higher mRAP was associated with increased age, male sex and a higher diastolic blood pressure but there was no difference in rates of heart failure types although the majority had preserved ejection fraction (51% vs 52%) with few having reduced ejection fraction (4.1% vs 8%). Right atrial (RA) dimensions, strain and ejection fraction had the strongest correlation with mRAP with moderate correlation to right ventricular (RV) measurements. Multivariable models contained simple RA dimensions (model 1), RA dimensions and strain (model 2), RA and RV dimensions (model 3) and RA end systolic volume corrected for body mass index and sex (model 4). All models had similar predictive capability (Figure 1 Panel A). RA end systolic volume was isolated in model 1 as the only variable required for mRAP prediction (coefficient = 0.06, p < 0.001). Using a threshold of mRAP >8mmHg ROC analysis demonstrated an area under the curve of 0.78 (95% Confidence interval 0.75 to 0.81) (Figure 1 Panel B). Worse WHO functional Class at time of assessment was significantly associated with increased CMR derived mRAP (P<0.001) (Figure 2). Conclusions mRAP can be estimated with moderate confidence from CMR RA end systolic volume. Limited available clinical data suggest functional status may be predicted by CMR derived mRAP. Further studies on wider populations are required to validate this model externally, refine it and determine its prognostic clinical significance.