(1) Background: To explore the effect of exercise on the formation and recovery of alcoholic liver disease (ALD) and whether the IL-6–p47phox oxidative–stress axis is involved in that process. (2) Methods: Firstly, 23 six-week-old male C57BL/6J mice were randomly divided into the Con group, ALD group, ALD + NOXI group, ALD + Ex group, and ALD + Ex + NOXI group. The Liber–DeCarli alcoholic liquid diet was used for 6 weeks to establish the ALD mice model, and the Con group was given the TP4030C control diet. The remaining groups were fed with the TP4030B alcoholic diet, and exercise intervention was started after the ALD model establishment and lasted for another 6 weeks, with or without administration of the NOX inhibitor apocynin by intraperitoneal injection on every exercise training day. Secondly, 28 mice were randomly divided into the Sed group, Eth group, Eth + Ex group and Eth + Ex + NOXI group. The Sed group was given the TP4030C control diet. The remaining groups were fed with the TP4030B alcoholic diet and exercise intervention was started synchronously combined with or without administration of intraperitoneal apocynin injections on every exercise training day for 5 weeks. After each individual experiment was accomplished, physiological assessment and biochemical analysis of blood and tissue samples were examined. (3) Results: The levels of TG in serum and IL-6 protein content in liver tissue in the ALD group were significantly increased compared to the Con group (p < 0.05); compared with ALD, p47phox expression in muscle was increased significantly in the ALD + NOXI group (p < 0.05), and TG in serum decreased in the ALD + Ex group (p < 0.05). TG in serum, AST/ALT ratio, and IL-6 content in both liver and muscle decreased (p < 0.05) in the ALD + Ex + NOXI group with MDA in muscle significantly increased (p < 0.01). The AST/ALT ratio, TG in serum, SOD in liver, and p47phox in both liver and muscle in the ALD + Ex + NOXI group were significantly decreased compared with the ALD + NOXI group (p < 0.01). Compared with the ALD + Ex group, the liver index and HDL-C levels in serum were decreased (p < 0.05) in the ALD + Ex + NOXI group. The degree of hepatocyte steatosis and inflammatory infiltration were ameliorated after exercise intervention. In the Eth group, the relative epididymal fat content, HDL-C level, and AST/ALT ratio were significantly decreased, and TG and gp91phox in liver were significantly higher than in the Sed group (p < 0.05, p < 0.01). Compared with the Eth group, the AST/ALT ratio, MDA in the liver, and NOX4 and p47phox protein expression in the liver were significantly increased, and body weight decreased significantly in the Eth + Ex group (p < 0.05, p < 0.01), as did TG in the liver and MDA in muscle. In the th + Ex + NOXI group, gp91phox expression in the liver and body weight were significantly decreased (p < 0.05, p < 0.01). In the Eth + Ex + NOXI group, the ratio of AST/ALT and MDA in muscle were increased when compared with the Eth + Ex group, and the protein expression of gp91phox and p47phox were much lower (p < 0.01). (4) Conclusions: 6 weeks of exercise intervention during the recovery phase of ALD ameliorates hepatocyte damage and dyslipidemia through the IL-6–p47phox oxidative–stress axis, and applying a NOX inhibitor in combination could optimize this. However, drinking alcohol during exercise exacerbates dyslipidemia and oxidative stress, with hepatocyte IL-6–p47phox downregulated.