While studies of bicyclist’s perceptions of crime and crash safety exist, it is also important to ask lower-income predominantly-minority residents what bicycle-route surface or context they perceive as safest from crime and crashes. With their insights, their chosen bike environments could be in engineering guidelines and built in their neighborhoods to improve residents’ health and lessen their risk of exposure to crime or crashing. This study involved two populations in Boston: (a) community-sense participants (eight groups-church/YMCA n = 116); and (b) street-sense participants (five groups-halfway house/homeless shelter/gang members n = 96). Participants ranked and described what they saw in 32 photographs of six types of bicycle environments. Quantitative data (Likert Scale 0–6 with 0 being low risk of crime/crash) involved regression analysis to test differences. Qualitative comments were categorized into 55 themes for surface or context and if high or low in association with crime or crashes. For crime, two-way cycle tracks had a significantly lower score (safest) than all others (2.35; p < 0.01) and share-use paths had a significantly higher score (least safe) (3.39; p < 0.01). For crashes, participants rated shared-use paths as safest (1.17) followed by two-way cycle tracks (1.68), one-way cycle tracks (2.95), bike lanes (4.06), sharrows (4.17), and roads (4.58), with a significant difference for any two groups (p < 0.01) except between bike lane and sharrow (p = 0.9). Street-sense participants ranked all, except shared-use paths, higher for crime and crash. For surface, wide two-way cycle tracks with freshly painted lines, stencils, and arrows were low risk for crime and a cycle track’s median, red color, stencils, and arrows low risk for crash. For context, clean signs, balconies, cafes, street lights, no cuts between buildings, and flowers were low risk for crime and witnesses, little traffic, and bike signals low risk for crash. As bicycle design guidelines and general Crime Perception Through Environmental Design (CPTED) principles do not include these details, perhaps new guidelines could be written.