Abstract Aims Non-invasive estimation of mean pulmonary artery pressure (mPAP) by cardiovascular magnetic resonance (CMR) four-dimensional (4D) flow analysis has shown excellent agreement with invasive right heart catheterization. However, clinical application is limited by relatively long scan times. Therefore, the aim of this study was to evaluate the accuracy and time reduction of compressed sensing (CS) accelerated acquisition for mPAP estimation. Methods and results Patients (n = 51) referred for clinical CMR at 1.5 T or 3 T underwent imaging with both a prototype CS-accelerated and a non-CS-accelerated flow sequence acquiring time-resolved multiple 2D slice phase-contrast three-directional velocity-encoded images covering the pulmonary artery. Prototype software was used for the blinded analysis of pulmonary artery (PA) vortex duration to estimate mPAP as previously validated. CS-accelerated and non-CS-accelerated acquisition showed increased mPAP in 22/51 (43%) and 24/51 (47%) patients, respectively. The mean bias for estimating mPAP between the two methods was 0.1 ± 1.9 mmHg and the intraclass correlation coefficient was 0.97 (95% confidence interval 0.94–0.98). Effective scan time was lower for the CS-accelerated acquisition (1 min 55 s ± 27 s vs. 9 min 6 s ± 2 min 20 s, P < 0.001, 79% reduction). Conclusions CS-accelerated CMR acquisition enables preserved accuracy for estimating mPAP compared to a non-CS-accelerated sequence, allowing for an average scan time of less than 2 min. CS-acceleration thereby increases the clinical utility of CMR 4D flow analysis to estimate mPAP.