Over the past few decades, work-zone crashes have been perceived as one of the pivotal concerns in improving roadway safety. Each year, traffic accidents in construction and maintenance zones result in hundreds of fatalities and thousands of injuries, and accordingly, many studies have investigated the statistical properties of work-zone crashes as part of efforts to reduce the numbers of fatalities and injuries. The primary goal of this study is to investigate the combined effects of various driving conditions on work-zone crash severity so that maximum speed in construction zone can be adjusted to meet the pre-defined safety. To this end, the paper first filters and clusters work-zone crash data and develops the marginal probability density functions of various exogenous and endogenous variables. In this study, vehicle speed and driving conditions (e.g., road surface, light and weather conditions) are identified as the key independent variables in crash severity analysis. Finally, the distributions of total work-zone crash cost conditioned on vehicle speed and various driving conditions are developed to specify the allowable speed limit of vehicles with the goal of reducing the crash costs. The proposed cost-based crash severity analysis can be further utilized in an online speed limit system considering various driving conditions.