A comprehensive before-and-after study evaluated the safety impacts associated with application of shoulder and centerline rumble strips on highways in British Columbia, Canada. Data were collected for three groups of sites: treatment group, comparison group, and reference group. The treatment group included 47 sites belonging to two highway classes: an undivided, rural two-lane arterial (RAU2) and a divided, rural four-lane freeway. A total of 225 sites were used to establish a comparison group for the treatment sites based on implementation year for the treatment site and the highway class. The comparison group was used to correct for the confounding factors of history and maturation. Six reference groups were used; they correspond to the two highway classes and three time periods (2000 to 2002, 2001 to 2003, and 2002 to 2004). Collision prediction models developed from the reference groups were used to correct for the regression to the mean and to account for the changes in traffic volumes in the before-and-after periods. Overall, the results showed that shoulder and centerline rumble strips can significantly reduce severe collisions and specific collision types: (a) the installation of rumble strips reduced all injury collisions by a statistically significant 18.0%; (b) shoulder rumble strips reduced off-road right collisions by a statistically significant 22.5%; and (c) centerline rumble strips (RAU2 sites) showed a statistically significant reduction of 29.3% in off-road left and head-on collisions. RAU2 sites with both centerline and shoulder rumble strips showed a statistically significant reduction of 21.4% in off-road right, off-road left, and head-on collisions combined.