Abstract: Breeding plot seeders are specialized machines used for field breeding experiments, and their working performance is considered essential to guaranteeing the reliability and accuracy of breeding test results. Nevertheless, several technical limitations are still encountered in currently available breeding plot seeders. Sufficient investigations into the continuous operation performance under different travel speeds and variable plot lengths have not been carried out, and low control precision for the starting seeding position and effective working length, accompanied by unsatisfactory sowing uniformity, are commonly observed. These drawbacks greatly hinder the popularization, application and intelligent upgrading of breeding plot seeding equipment. To tackle the above-mentioned problems, an automatic control system for breeding plot seeders is proposed in this research on the basis of prescription map-driven parameter identification and precise regulation of seeding operational parameters, so that non-stop continuous seeding operations can be achieved for breeding plots with variable lengths. In compliance with the agronomic requirements of plot seeding for breeding trials, an annular chain-type seed feeding device is designed, by means of which automatic seed supply for nine separate plots can be completed within one single working pass. Core functional components including the seed dropping, cone compartment tray and centrifugal seed distribution are analyzed and designed, and their critical structural parameters are determined. Built upon operation prescription maps, the automatic control framework is established, and precise control models for automatic seed feeding, automatic seed dropping, seed transport rotating speed of the cone compartment tray and rotating speed of centrifugal seed distribution are constructed. Combined with machine travel speed, real-time position information and parameters extracted from operation prescription maps, real-time matching and dynamic regulation of operational parameters under different plot lengths are realized, and fully automated seeding operations are fulfilled. The starting seeding position is affected by multiple factors such as control system response delay, actuator execution error, seed falling time lag and machine travel speed. For this reason, a deviation model for the initial seeding position of breeding plots is developed. When the machine travel speed is less than or equal to 1.74 km/h, the actual starting seeding position is found to lag behind the preset position, and lag compensation is required. When the machine travel speed exceeds 1.74 km/h, the actual starting seeding position passes the preset position in advance; therefore, advance prediction and prescription map recognition are adopted to trigger seeding actions ahead of schedule. Comparative seeding performance tests are conducted. For plot lengths of 2.91 m, 5.85 m and 8.99 m, the deviation between the actual seeding length obtained with the proposed automatic control system and the preset value is kept below 0.14 m under various travel speeds. The variation coefficients for the consistency of seeding amount per row are measured as 6.33%, 6.08% and 5.92%, respectively, while the coefficient of variation of in-row sowing uniformity reach 20.88%, 19.74% and 19.57%. In contrast, under the traditional mechanical driving mode, the average deviation between actual seeding length and preset value is greater than 0.5 m, the variation coefficients for the consistency of seeding amount per row are 7.57%, 6.72% and 6.77%, and the coefficient of variation of in-row sowing uniformity are 24.87%, 22.97% and 21.94%. It is verified that the seeding performance of the breeding plot seeder equipped with the developed automatic control system is significantly improved compared with the traditional mechanical driving method. Continuous variable-length plot seeding experiments based on plot prescription maps are further implemented. Nine plots with a length of 3 m, six plots with a length of 6 m and six plots with a length of 9 m are arranged, and the interval widths between adjacent plots are set to 1 m, 0.7 m and 0.4 m, respectively. The test results show that the measurement errors of actual plot length and interval zone width after seedling emergence are both less than 0.10 m, and the stability coefficients are higher than 98% and 90%, respectively. These indicators satisfy the agronomic requirements of breeding plot seeding. The research outcomes can provide valuable references for the structural optimization design and intelligent control system development of breeding plot seeders.