Red fruit oil (RFO) is a high-value oil that contains functional compounds, mainly phenolic compounds, providing antioxidant activity. Therefore, an optimal extraction method is essential to recover the RFO and phenolic compounds simultaneously. This research aimed to optimize the ultrasound-assisted extraction (UAE) for oil from red fruit using the Box-Behnken design combined with response surface methodology. The studied UAE factors, including sample-to-solvent ratio (1:3, 1:2, and 1:1 g mL−1), extraction temperature (60, 75, and 90 °C), and pulse duty-cycle (0.20, 0.50, and 0.80 s−1). Analysis of variance revealed that the three studied factors significantly influenced the recovered RFO, while the level of total phenolic compounds in the extracts was defined merely by extraction temperature (p < 0.05). These significant factors were then included in the optimization models (R2 > 0.99, lack-of-fit p > 0.05). The proposed UAE setting by the multiresponse optimization was an extraction temperature of 67 °C, a pulse duty-cycle of 0.50 s−1, and a sample-to-solvent ratio of 1:2.5 g mL−1. Subsequently, the extraction kinetic was evaluated, confirming full recovery at 60 min of extraction time. The developed method was then applied to extract six red fruit clones. Mbarugum clones provided high RFO recovery (9.60%), with an uppermost total phenolic compound of (42.63 mg GAE g−1) among the six red fruit clones. Additionally, the resulting RFO showed eminent antioxidant activities, indicated by excellent values of IC50 DPPH (37.69 mg L−1), IC50 FIC (30.43 mg L−1), FRAP reducing power (63.55 mg AAEA g−1), and IC50 ABTS (93.88 mg L−1). In contrast with a wet rendering method, UAE enhanced the RFO recovery by 53.02%, resulting in a higher level of total phenolic compounds. Henceforth, the proposed UAE method is a promising technique to substitute conventional oil production in the food and pharmaceutical industries.