Abstract Artificial insemination (AI) with fresh semen is the primary breeding tool in the swine industry. However, the heterogenous semen doses create challenges in boar studs due to the lack of consistent assessment criteria. Seminal plasma (SP) surrounding spermatozoa is a great source of reliable non-invasive biomarkers of sperm quality. It contains a wide array of bioactive molecules, whose dynamic compositions in response to extrinsic and intrinsic factors influence sperm motility and morphology. Therefore, we hypothesize that the miRNA content of extracellular vesicles (EVs) found in SP may reflect semen quality. Our study aimed to characterize poor vs. good quality semen and compare miRNA profiles of SP-derived EVs. Raw semen was collected (n = 75) from sexually mature Duroc boars at a commercial stud (Prestage Farms, MS) for 8 weeks. Semen samples were subjected to sperm motility and morphology analyses and subsequently classified as poor (n = 38) or good (n = 37) quality based on assessment criteria cut-offs (< 70% and ≥70%, respectively). Thereafter, raw semen samples were subjected to serial centrifugation to collect spermatozoa and SP. Samples were stored at -80oC until analyses. A subset of each frozen-thawed SP was used for NIRS absorbance analysis (ASD FieldSpec3 portable spectrometer – 1mm quartz cuvettes; 750-2500 nm). Other subsets of frozen-thawed SP were ultra-centrifugated to isolate EVs (SP-EV), followed by confirmations through transmission electron microscopy (TEM), dynamic light scattering (DLS), and immuno-western blotting. Five pools of three individual SP-EV samples, corresponding to extremely poor and extremely good quality semen, were constituted for total exosome RNA isolation, small RNA sequencing, miRNA detection, and differential miRNA expression. Messenger RNA targets of miRNA were searched and subjected to functional bioinformatic analyses using various software algorithms. Significant differences were set for P < 0.05. Poor and good quality samples had different spectrum profiles following NIRS analyses. Changes were explicitly observed in water structures at coordinates C1, C5, and C12 (nm), corresponding to water associations with chaotropic solutes (water asymmetrical stretching vibrations-ν3), free water, and kosmotropic solutes (increased H-bonding numbers and strength), respectively. The NIRS-LDA data contrasted two distinctive poor and good quality SP samples. All boar SP-EV were confirmed: round spherical structure; 30 to 400nm of diameter; and CD9+, CD63+, and CD81+ membrane markers. Totals of 281 and 271 miRNAs were identified in poor and good groups, respectively, and 15 were differentially expressed (DE). Only three DE miRNAs (ssc-miR-205, ssc-miR-493-5p, and ssc-miR-378b-3p) permitted targeting of 109 mRNA, which functional analyses indicated associations with nuclear and cytosol localization, as well as with functions, i.e., acetylation, Ubl conjugation, protein kinase binding, potentially altering sperm function. In conclusion, the study indicates that the NIRS tool can distinguish between good and poor semen quality, and the miRNA SP-EV content may reflect semen quality.