Contents lists available at ScienceDirect Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap Bike lanes next to on-street parallel parking Paul Schimek Effective Data Associates, 50 Saint Rose Street, Boston, MA, 02130, United States ARTICLE INFO Keywords: Bicycle Bicycling Bike lanes Separated bike lanes Cycle tracks On-street parking ABSTRACT For decades it has been the conventional wisdom that crashes involving bicyclists and opening car doors are rare. This belief is based on motor vehicle crash reports, but these reports generally exclude this crash type by de- finition. More complete sources show that dooring crashes are one of the most common causes of urban bicycle- motor vehicle collisions, accounting for 12%–27% of the total. This paper reviews all available studies of bicyclist position in bike lanes adjacent to on-street parking. With bike lanes meeting current minimum standards, almost all bicyclists were observed riding within range of opening doors. However, when an additional three or four feet is provided between the bike lane and parked cars, hardly any bicyclists are observed in the door zone. All of the design guides recently developed in North America for separated bike lanes include a buffer to account for the door zone when the bike lane is placed between on-street parallel parking and the curb. However, only the Ontario design guide has a similar requirement for standard bike lanes. The buffer require- ment for standard bike lanes adjacent to on-street parking should be incorporated into all design guidance. When there is not room for this necessary buffer, an alternative is to place a shared lane marking in the center of the travel lane, which encourages bicyclists to ride outside the door zone. Increasing the number of bicyclists who ride outside of the door zone may require lowering speed limits and repealing laws that create a pre- sumption that bicyclists must always keep to the right of the travel lane. 1. Introduction There has been a large increase in the number of marked bike lanes in North American cities in the past two decades. Many of these lanes have been added in older urban areas, where arterials often have on- street parking. One of the motivations for marking bike lanes is to make bicyclists feel welcome on city streets. However, government agencies and bicyclist organizations routinely warn bicyclists about the danger of suddenly opened door of a parked car – a problem known as “dooring”– even when bicyclists are using bicycle lanes. How frequently do bicyclists strike the opened door of a parked car? A recent paper argues that “past studies have shown that dooring cra- shes are a rare form of bicycle crash and are not relatively dangerous” (Ferenchak and Marshall, 2016). Is this contention supported by available data? This article reviews (a) the available data on dooring crashes to determine their prevalence; (b) studies that include ob- servational data on bicyclist position with respect to on-street parking in the presence of different lane widths and markings; and (c) design guidance for separated and ordinary bike lanes to determine how they account for the “door zone.” (This article is concerned only with parallel on-street parking. Angle on-street parking does not present a dooring hazard for bicyclists, although it does present a backing hazard, parti- cularly with the more common back-out design.) 2. Prevalence of collisions with car doors The main source of U.S. data on bicyclist crashes is the National Highway Traffic Safety Administration, which provides both a nation- ally representative sample of police-reported crashes and a complete inventory of road fatalities. However, both databases are restricted to crashes involving a motor vehicle in transport. Bicycles are not “motor vehicles” and parked motor vehicles are not “in transport.” Therefore, dooring collisions are excluded by definition from these national da- tabases, as well as from U.S. state crash databases. This exclusion may not be clear to data users, since in some datasets a few of these crash types are included, possibly inadvertently. For example, the NHTSA General Estimates System includes the Pedestrian and Bicycle Crash Analysis Tool (PBCAT) code for “Bicyclist Overtaking - Extended Door”, and two out of the 3437 bicycle-car collisions in the 2015 sample (less than 0.1%) were assigned this code (Harkey et al., 1996). North Car- olina is the only U.S. state that routinely codes police-reported crashes using the PBCAT system. Of the more than 17,000 North Carolina https://doi.org/10.1016/j.aap.2018.08.002 Received 5 April 2018; Received in revised form 2 August 2018; Accepted 2 August 2018 E-mail address: paul.schimek@effectivedataassociates.com. Accident Analysis and Prevention 120 (2018) 74–82 0001-4575/ © 2018 Published by Elsevier Ltd. T