Purpose <p>Information is limited on prediction models for length of stay (LOS), observed clinical reasons for prolonged LOS, and the association of prolonged LOS with outcome metrics after elective lumbar decompressive surgery.</p> Methods <p>We analysed prospectively collected data on 3132 cases with or without instrumented fusion. A prediction model for LOS in hospital was developed with 2008–2022 data and validated on 2023 data. Validation data was used to evaluate LOS model R² and area under the curve (AUC) to discriminate to prolonged LOS. Observed reasons for prolonged LOS were categorised. Prolonged LOS association with unplanned 30-day readmission was evaluated by absolute risk increase (ARI), and association with outpatient physiotherapy session frequency and 12-month Oswestry Disability Index (ODI) by mean difference.</p> Results <p>All-procedure LOS reduced from 3.0 days in 2008 to 1.5 days in 2023. Notable preoperative LOS predictors were fusion, multilevel surgery, mobility aided, ≥ 80 years, ASA 3, diabetes, and female. LOS model R² was 0.19 (95% CI 0.12–0.26) and AUC was 0.64 (95% CI 0.53–0.75). Prolonged LOS occurred in 11.3% (355/3132) with the observed reason for delayed discharge identified as dural tear (2.7%), difficulty to mobilise (1.5%), pain control (1.2%), patient expectation (1.2%), gastrointestinal (0.7%), and miscellaneous (4.1%). Prolonged LOS was associated with unplanned 30-day readmission (ARI 3.5%, <i>p</i> &lt; 0.001), more outpatient physiotherapy (Δ 0.8 sessions, <i>p</i> &lt; 0.001), and poorer 12-month disability (Δ ODI 5.6, <i>p</i> &lt; 0.001).</p> Conclusion <p>LOS was reduced with no reciprocal increase in unplanned 30-day readmission. Safe reduction of LOS requires enhanced recovery after surgery multidisciplinary pathways that focus on LOS risk factors and perioperative complications. Prolonged LOS is associated with poorer outcomes.</p>