This paper investigates the relationship between medical treatment costs and the length of hospital stays resulting from motorcycle crashes involving the elderly. The World Health Organization defines 'elderly' as people more than 65 years old. The sample for this study consisted of data for the year 2007 collected by the Bureau of National Health Insurance, Taiwan. We develop models for predicting medical costs and the length of hospital stays based on diagnosis, hospital and user types. The seemingly unrelated regression equation (SURE) model was applied first to investigate the relationship between medical costs and the length of hospital stays. The SURE model shows that the type of injury (e.g. head injury) is statistically significant and has positive effects on medical costs for motorcycle crashes involving the elderly in Taiwan. Due to the statistical insignificance of the dependency between medical costs and length of hospital stays, two separate simple linear regression models were subsequently estimated. For motorcycle crashes, patients over 80 years old had the highest medical costs. The findings reinforce the need for transportation authorities to focus on preventing certain types of injuries that are particularly serious and costly for the elderly in Taiwan.