In the last decade, naturalistic driving studies (NDSs) have given researchers an unprecedented way to study the behavior of drivers through the deployment of, and capturing the data from, on-board vehicle sensors and cameras. The ability to determine the dominant driving risk facto rs can play an essenti al role in shaping transportation policies and education programs for drivers. This paper presents a cohort study statistical analysis to determine the risks associated with traffic and road surface conditions, quanti fied in terms of crash and near crash events. Two risk quantification measures, odds ratio (OR) and relative risk (RR, are utilized to signify the associated risk. For this research we used the 100-CAR data set, with a total of 829 crash and near crash and 19616 baseline events, which are driving events captured randomly in normal driving episodes. In the 100-CAR data set, traffic density is divid ed into six levels according to the traffic flow con dition. Similarly, road su rface condition is divided into four categories. To quantify the statistical significance of the results, measures such as the p-value are employed. The results show that icy roads with level-of-service (LOS) A, wet roads with LOS D, and dry roads with LOS D have the highest risk for crashes and near crashes. These results are proven to be of statistical significance.