Maintenance and repair of bridges represent significant costs in provincial and municipal government budgets. Prediction of bridge conditions can help managers in annual cost estimating and budget allocation. To assess Bridge Condition Index (BCI), each bridge component must be inspected every two years, tested if it is required, and rated. Bridge condition can be affected over time by different attributes such as material, structure, location, and use. This paper presents a study conducted to model and predict BCI based on a historical dataset of 2803 bridges in Ontario from 2000 to 2014. The paper describes the work related to data collection, cleaning and transformation. In addition, a comparison of the cross-validation performance of alternative BCI prediction models is presented and discussed.