AbstractThis paper provides an observational before/after evaluation of the safety effects of edge-lane road (ELR) (also known as advisory bike lanes or advisory shoulders) installations in the United States. An ELR is a class of roadway that supports two-way automobile traffic within a single center lane and vulnerable road users (VRUs), i.e., bicyclists or pedestrians, in the edge lanes on either side. The use of a single lane by automobile users traveling in both directions is often a cause of potential safety concerns among the general public. This study employs a project-level empirical Bayes (EB) approach to before/after safety analysis for all US ELR sites where requisite crash data and other relevant characteristics were available. The analysis at 11 sites was performed with 8 years of crash data and more than approximately 60 million motor vehicle trips. Project-level EB analysis based on safety performance functions (SPFs) showed 8 of 11 ELR sites experienced a reduction in crash experience since installation. There was a ∼44% reduction in crashes among all sites compared to the expected crashes on the traditional two-lane two-way design that existed before ELR installation. This estimation assumed the calibration factor for all SPFs used in the EB analysis was 1.0. Because of the geographical spread of the ELRs being analyzed in this study, estimation of specific calibration factors for all 11 sites was beyond the scope of this work. To address this limitation, we conservatively assumed the SPF calibration factors to be 0.50 for locations for which calibration factors were not available. This assumption was made for 7 of the 11 sites. Even with this conservative assumption, the ELRs were estimated to have a ∼36% crash reduction. The results from this comprehensive evaluation of existing US ELRs should alleviate concerns the general public often has about the safety of ELRs for automobile users.