This study analyzes the traffic flow conditions that affect crash occurrence on freeway ramps by type (on- or off-ramps) and configurations (diamond, loop, etc.). The study used the 5-min traffic flow data before the crash obtained from loop detectors to identify the traffic conditions contributing to crashes on ramps. With 5 years of ramp crash data on the Interstate 4 freeway in Orlando, Florida, a two-level nested logit model was developed to estimate the probabilities of crash occurrence for different ramp types and configurations. In the comparison of two nest structures, the traffic flow parameters contributing to crash occurrence greatly differed between on-ramps and off-ramps. The results of the model estimation suggested that the main-line speeds immediately upstream and downstream of ramps and the volume on ramps were correlated to crash occurrence on ramps. It is recommended that these traffic flow parameters be monitored in real time to detect elevated risk in traffic conditions on ramps within on-line systems for freeway traffic management.