Abstract: A seed-metering device is confined to the seed-airflow-mechanical multi-field coupling mechanism during high-speed operation of the maize delta-row dense planting planter. It is also unclear on the seeds movement and airflow distribution, leading to the low performance of the device. This study aims to explore the overall structure and working principle of the air-pressure high-speed precision seed-metering device. Theoretical analysis, simulation, bench testing, and field trials were also carried out to investigate the influence of the key parameters on the performance of the device. Theoretical models were then established for the seed-filling and seed-cleaning of the device. DEM-CFD coupling simulation was used to explore the flow field of the device chamber and the shape holes. A systematic analysis was made on the variation trend of the average drag force and coefficient of variation of speed of the qualified delta-row group seeds in the seed-cleaning zone at different seed-cleaning angles. The specific motion was observed for the big-rounded, big-flat, small-rounded, and small-flat seeds in the device. Some indicators were defined for the quality evaluation on the delta-row sowing of the device, including the qualified index of delta-row, qualified index of row spacing, projection spacing qualifying coefficient of variation, and row spacing qualifying coefficient of variation. Full-factor bench tests were conducted with the seed-cleaning angle, chamber inlet pressure, and operating speed as the experimental factors. While the high-speed sowing field tests were conducted on the maize delta-row dense planting in different tillage patterns. Simulation tests show that the coefficients of variation of pressure and flow velocity were less than 2.9% and 4.8%, respectively, when the chamber inlet pressure was in the range of 3.6-4.2 kPa. There was a more uniform and stable distribution of the flow field in the device. The average drag force of the seeds in the qualified delta-row group was higher than 0.06 N in the range of the seed-cleaning angle of 3.0°-4.0°, where the coefficient of variation of speed was lower than 5.29%, indicating more stable seed-cleaning. In the motion of each type of seed, the average drag force of the big-rounded seeds was 0.0793 N from the stable seed-filling to the seed-unloading point, indicating the more suitable for the seed type of sowing. The bench tests showed that the qualified index of the delta-row of the device was more than 70%, and the projection spacing qualifying coefficient of variation was less than 12% at the seed-cleaning angle of 3.0°-4.0°, chamber inlet pressure of 3.6-4.2 kPa, and operating speed of 12-16 km/h. Field test results show that the qualified indexes of the delta-row of the device and row spacing were more than 72% and 85%, respectively. The projection spacing qualifying coefficient of variation was less than 14%, with the seed-cleaning angle of 4°, the chamber inlet pressure of 4.2 kPa, and the operating speed of 12-16 km/h. The row spacing qualifying coefficient of variation was less than 8% under various tillage patterns. The device can be expected to maintain the high quality of delta-row sowing in both flat breaking and ridge plowing modes, fully meeting the requirements for the precision sowing of maize. This finding can provide a basis to further improve the performance of the device.