This study examined the integration of downloaded instructional videos in learning Agriculture and the test scores of Grade 9 students at Mahinog National High School II during School Year 2025-2026. A quasi-experimental pretest-posttest design was used with two intact heterogeneous classes: 36 learners in the control group and 36 learners in the experimental group. The control group received lecture-based instruction without digital aids, while the experimental group received multimedia presentations and downloaded video demonstrations on second-quarter Agriculture lessons. A 30-item teacher-made multiple-choice test was administered before and after the three-week intervention. Frequency, percentage, mean, standard deviation, independent-samples t-test, and regression analysis were used. The experimental group obtained higher mean scores than the control group in both the pretest (M = 7.61, SD = 1.49 versus M = 7.34, SD = 1.94) and posttest (M = 8.44, SD = 1.04 versus M = 8.20, SD = 1.59). However, the direct comparison between groups was not statistically significant, t(70) = 0.74, p = .46. The reported regression model was significant, F(3, 68) = 32.66, p < .001, and explained 59% of the variance in posttest scores (R² = .59). The reported interaction term was significant (p = .03), suggesting that the relationship between pretest and posttest performance varied across instructional conditions. The findings indicate that downloaded videos may support learning consistency and modest score gains, but the intervention did not produce a statistically significant direct advantage over lecture-based instruction. Further studies with clearer regression specifications, larger samples, and longer interventions are recommended.