Background: Workers' compensation data are an important source for evaluating costs associated with construction injuries. Methods: We describe the characteristics of injured construction workers filing claims in Illinois between 2000 and 2005 and the factors associated with compensation costs using a robust regression model. Results: In the final multivariable model, the cumulative percent temporary and permanent disability-measures of severity of injury-explained 38.7% of the variance of cost. Attorney costs explained only 0.3% of the variance of the dependent variable. Discussion: The model used in this study clearly indicated that percent disability was the most important determinant of cost, although the method and uniformity of percent impairment allocation could be better elucidated. There is a need to integrate analytical methods that are suitable for skewed data when analyzing claim costs.