OBJECTIVES: Early readmissions place a considerable, yet potentially avoidable, strain on tertiary healthcare resources. However, accurately identifying subgroups of patients who are both at risk of early readmissions but yet stand to benefit from preventive interventions, still remains a challenge. We therefore aimed to develop and externally validate a 15-day readmission risk prediction model to address this unmet need.  METHODS: A case-control analysis was performed on patients discharged from Khoo Teck Puat Hospital between January and March 2013. Cases were patients who experienced unplanned readmissions within 15 days of discharge. Multivariate logistic regression was used to derive a classification model. Temporal and geographical validation were performed between August and September 2015 at Khoo Teck Puat Hospital and Singapore General Hospital, respectively. Model discrimination was ascertained using the receiver operating characteristic (ROC) curve.  RESULTS: Number of medications prescribed at discharge [odds ratio (OR): 1.06, 95% confidence interval (CI):1.01-1.12], discharge to a nursing home (OR: 3.57, 95% CI:1.57 – 8.34), discharging against medical advice (AOR: 5.05, 95% CI:1.20 – 21.23), age (OR: 1.02, 95% CI:1.01–1.03) and a prior diagnosis of malignancy (OR: 3.37, 95% CI:1.16 - 9.80), peptic ulcer disease (OR:3.05, 95% CI:1.12-8.26), chronic obstructive pulmonary disease (OR: 3.16, 95%CI:1.12 - 8.05) and anemia (OR: 2.08, 95% CI: 1.15 - 8.05) independently predicted 15-day unplanned readmission risk. This model has a sensitivity of 55.6%, specificity of 70.7% and ROC of 0.65 on temporal validation [n=101, 9 readmitted (8.9%)]. On geographical validation, model displays a sensitivity of 57.5%, specificity of 60.9% and an ROC of 0.64 [n=520, 73 readmitted (14.0%)]. CONCLUSIONS: The proposed 15-day readmission risk prediction model showed reasonable discriminative power using information that is easily available during discharge. The model can assist hospitalists in selectively administering preventive interventions at patients who are most likely to benefit from these initiatives.