Aim Careful resource allocation is critical in health care. This study aims to identify preoperative factors predicting the need for postoperative respiratory support after tonsillectomy, to optimize ICU bed utilization and reduce unplanned admissions. Methods Retrospective review of patients under 18 who underwent tonsillectomy at a tertiary paediatric hospital between July 2022 and February 2023. Variables extracted included age, gender, comorbidities, preoperative overnight oximetry, preoperative CPAP use, and polysomnography respiratory disturbance index. Postoperative respiratory interventions in the PACU and on the ward were recorded. Ordinal logistic regression was performed to identify predictors of ward respiratory interventions. Interventions were then dichotomized into low-level (no intervention or low-flow oxygen) and high-level (high-flow oxygen, CPAP, intubation) for binomial logistic regression. Analyses were conducted with and without ONO to identify potential surrogate predictors when oximetry is unavailable. Results Two hundred patients analysed. Preoperative CPAP predicted escalated postoperative support (OR ~13, p = 0.004), as did severe OSA on ONO (OR 2.7, p = 0.019) and obesity (OR ~13, p = 0.005). Neurologic comorbidities trended toward higher intervention needs (OR 3.6, p = 0.096). Gender, history of cardiac/airway anomalies, prematurity, and Trisomy 21, were not significant. Excluding ONO revealed additional predictors (neurologic abnormalities, age, and PACU events) without evidence of mediation after conducting mediation analysis. Only 6 required high level care, and they were predicted preoperatively and intraoperatively. No patients required escalation of care from PACU. Conclusion Preoperative CPAP use, obesity, and severe OSA are strong predictors of postoperative respiratory support. Where ONO is unavailable, PACU events, neurologic comorbidities, and obesity can guide risk stratification. Many patients can have their care de-escalated.