One of the main goals in incident management is to reduce delay and queuing caused by an incident. Delay and queuing are among other things dependent on the road capacity, so an accurate estimate of the capacity reduction can potentially lead to improved efficiency of incident management and delay reduction. In this study a new method based on the slope of the cumulative counts is proposed to estimate capacity reductions at the incident site. The method is applied to real world incident scenarios on four lane freeways in Maryland and Northern Virginia. Thirty-two incidents from Maryland and nineteen cases from Northern Virginia meet the requirements to apply the proposed methodology. Results indicate that except for shoulder accident cases, the estimated available capacity ratios during incidents are significantly different from those reported in HCM. It is found that in cases where one driving lane is blocked, the available capacity reduces to 70%. This value reduces to 36% and 17% for two lanes closure and three lanes closure scenarios, respectively. Moreover, it is found that besides less available number of lanes, less efficient use of driving lanes due to changes in driver behavior and distractions is another contributing factor to capacity reduction.