This study presents a new method of analyzing whether a vehicle is in a dangerous driving condition while travelling along a highway. We use data from various sensors installed on the vehicle, which represent the driving attributes associated with a particular driving scenario, as inputs to our system. However, as some of these sensors may be dependent on each other, using redundant attributes to analyze the driving conditions can become very time consuming. Therefore, our dangerous driving condition analysis system (DDCAS) first selects discriminative attributes using a fuzzy rough sets technique. Next, based on these selected attributes a set of association rules is constructed, which is then used to infer whether a driving condition is hazardous or safe. If the driver is detected to be in a dangerous driving condition, the DDCAS outputs warning messages to the driver in an attempt to reduce the likelihood of an accident. This paper outlines experiments, which were conducted with a simulated system. In the future, we will install the DDCAS onto a real vehicle, with the aim of reducing the number of real accidents caused due to dangerous driving conditions.