Background <p>A validated prediction model and mediation analysis for postoperative nausea (PON) are essential to guide targeted prevention and treatment in metabolic bariatric surgery.</p> Methods <p>This study analyzed 2,331 anonymized metabolic bariatric surgery cases from the PAIN OUT database. PON was assessed on the first postoperative day using a standardized questionnaire. Clinically significant PON was defined as a numeric rating scale ≥ 4, categorizing participants into PON_CS (scores 4–10) and Control (scores 0–3) groups.</p> Results <p>A total of 1,486 patients (63.7%) were assigned to the Control group and 845 (36.3%) to the PON_CS group. Among 30 candidate predictors, the following were identified as independent predictive factors for PON: age (OR [95% CI]: 0.979 [0.970–0.987]), female gender (2.159 [1.738–2.682]), hypertension (0.785 [0.635–0.972]), diabetes (0.572 [0.420–0.779]), cancer (0.474 [0.259–0.869]), affective disorders (0.616 [0.497–0.763]), chronic pain (1.272 [1.011–1.600]), preoperative non-opioids (0.619 [0.430–0.890]), sleeve gastrectomy (1.710 [1.389–2.106]), and three postoperative interventions administered in the PACU—non-opioids (1.482 [1.183–1.855]), opioids (1.536 [1.219–1.936]), and regional nerve blocks (0.413 [0.274–0.622]). The model showed good discrimination (AUC = 0.709; 95% CI 0.687–0.730) and calibration. Internal validation yielded an optimism-corrected AUC of 0.708 (95% CI 0.687–0.730) and a calibration slope of 0.955. PACU-administered non-opioids, opioids, and regional nerve blocks mediated the effects of preoperative non-opioids use and chronic pain on the PON development.</p> Conclusion <p>The novel PON prediction model shows favorable performance in metabolic bariatric surgery patients, with PACU-administered non-opioids, opioids, and regional nerve blocks identified as modifiable mediators linking preoperative non-opioids use and chronic pain to PON development.</p>