ObjectiveReproductive hormones are a traditional good method to evaluate spermatogenesis but might not accurately represent local spermatogenesis. To find a more accurate method, seminal reproductive hormones were studied.MethodsA bidirectional cohort study was performed. A total of 126 infertile men from 2018 to 2019 were retrospectively analyzed. They were divided into nonobstructive azoospermia (NOA), oligozoospermia (OLZ) and normal (NOR) groups. A prospective study was conducted on patients in the NOA and OLZ groups for 2 years. Microscopic testicular sperm extraction was performed for NOA patients, who were divided into a focal spermatogenesis group (FS) and an idiopathic azoospermia group (IA). Drug treatment was for OLZ patients, who were divided into a valid group (VA) and an invalid group (IN). The differences in sperm parameters and reproductive hormones were compared. ANOSIM analysis was used between and within groups. Pearson correlation analysis, CO inertia analysis and Proctor’s analysis were for relationships. ROC curve for the specificity and sensitivity. Time series analysis was for the trends between hormones and time.ResultsThe b-FSH, b-LH, s-T and ΔT in the NOA group were significantly higher than those in the OLZ and NOR groups. However, the s-FSH, s-E2, s-P, ΔFSH, ΔLH, ΔP and ΔE2 were lower. Thirty-one NOA patients underwent MTSE, of whom 12 had sperm (FS) and 19 had no sperm (IA). The s-FSH and s-E2 of the FS group were higher than those of the IA group. Twenty-six OLZ patients completed 30 days of treatment, of which 11 had an improved sperm count (VA) and 15 had no (IN). The ΔT of the VA group was higher than that of the IN group. After follow-up for 2 years, 18 patients’ results showed that b-FSH, b-LH and s-T were different over time, with delays of 19, 3 and -19 days. SC is closely related to pH, s-FSH, s-LH, s-E2, s-P, s-T, b-FSH, b-LH, ΔFSH, ΔLH, ΔP, ΔE2 and ΔT. There were complex common trends and relationships between different kinds of hormones. s-FSH, s-LH, s-E2, s-P, s-T, b-FSH and b-LH were useful to judge spermatogenesis, of which s-T, b-FSH and b-LH were more sensitive. If s-T, b-FSH and b-LH reached 64.4, 9.4 and 4.7, respectively, their prediction performance was the strongest.ConclusionSeminal testosterone is sensitive for judging local spermatogenesis in nonobstructive azoospermia patients, which may be the direction of local spermatogenesis in nonobstructive azoospermia.Clinical trial registrationhttp://www.chictr.org.cn/index.aspx, identifier ChiCTR2200060463.