Research on secondary crashes has been limited, mainly due to the poor quality of incident data and related traffic data that are necessary for secondary incident analysis. In recent years, traffic management agencies have begun the implementation and maintenance of detailed databases that can be used for detailed analysis of incidents including secondary crashes. This paper utilizes such a database in an effort to examine the relationships between the primary incident characteristics and the likelihood and severity of secondary crashes. Summary statistics and logistic regression analyses were applied to understand these relationships. The derived regression model indicates that five factors have significant effects on the likelihood of incident occurrence. These factors are: the number of vehicles involved in the primary incident, the number of lanes at the primary incident site, the primary incident duration, time-of-day of incident occurrence, and if vehicle rollover occurs during the primary incident. Secondary crash severity analysis confirms that secondary crashes are usually much less severe than other crashes, which could be an important factor when conducting benefit-cost analysis for incident management. The results show that incident visibility and the lane blockage durations of the primary incidents are significant contributing factors for determining the severity of secondary crashes.