In this research we study the effect of CMV drivers' medical conditions on crash severity. To our knowledge, no study has ever isolated this effect. The severity of a crash is measured by the total number of victims (injured and dead). We estimate count data regression models which incorporate, simultaneously, information on drivers' characteristics, crash circumstances and health status. We emphasize the fact that these variables have to be controlled adequately in order to isolate the effects of health status on crash severity. Our results show that truck drivers with binocular vision problems and bus drivers with hypertension are involved in more severe crashes than healthy drivers. No other medical condition considered in this study has a significant effect on crash severity. Many variables describing crash circumstances are also significant.