OBJECTIVE: To model and predict hospitalisation costs after first acute coronary syndrome (ACS) based on age, sex, social class, previous hospitalisation for diabetes, COAD and renal diseases. METHODS: The Cramer-Lundberg insurance model was adapted to predict the hosptalisation costs within a given period. Three important factors, costs per hospitalisation, time to next hospitalisation and time to death were modelled. The costs were modelled by a gamma regression model using the GEE approach to accommodate intra-patient correlation. Parametric survival models were used for easy prediction. A Weibull regression was used for the recurrence of hospitalisations. Future costs can be predicted by combining the models and using simulation. RESULTS: Average costs increase with patient's age. Compared with patients over 80, the costs per hospital admission of those under 40 was only 44%. Previous hospitalsation for diabetes, COAD and renal diseases increases the costs by 25%, 12% and 12% respectively. Younger patients were more likely to have a further hospitalisation. The log-RR for those under 40 is 0.36 (95% CI=0.24, 0.48) compared with those over 80. On average, each previous ACS increases the log-risk by 0.27 (95% CI=0.23, 0.31). Previous hospital admission for diabetes and renal diseases also indicated higher risk with log-risk 0.34 (95% CI=0.21, 0.47) and 0.53 (95%CI=0.28, 0.77) respectively. Younger patients had lower mortality with log-RR for those under 40 is -2.78 (95%CI= -3.05, -2.50), compared with those aged over 80. Previous hospital admissions for diabetes, COAD and renal disease also indicated higher mortality with log-RR 0.22 (95%CI= 0.08, 0.36), 0.38 (95%CI= 0.27, 0.49) and 0.86 (95%CI= 0.68, 1.03) respectively. CONCLUSION: The combined model procedure provided a flexible approach to analyse and predict hospitalisation costs. Some factors may affect the costs in several ways. This approach explored the role of the factors in predicting individual costs.