There is growing evidence that crime is strongly concentrated in micro-geographic hot spots, a fact that has led to the wide-scale use of hot spots policing programs. Such programs are ordinarily focused on deterrence due to police presence, or other law enforcement interventions at hot spots. However, preliminary basic research studies suggest that informal social controls may also be an important mechanism for crime reduction on high crime streets. Such research has been hindered by a lack of data on social and attitudinal characteristics of residents, and the fact that census information is not available at the micro-geographic level. Our study, conducted in Baltimore, MD, on a sample of 449 residential street segments, overcame these limitations by collecting an average of eight surveys (N = 3738), as well as physical observations, on segments studied. This unique primary data collection allowed us to develop the first direct indicators of collective efficacy at the micro-geographic level, as well as a wide array of indicators of other possible risk and protective factors for crime. Using multilevel negative binomial regression models, we also take into account community-level influences, and oversample crime hot spots to allow for robust comparisons across streets. Our study confirms the importance of opportunity features of streets such as population size and business activity in understanding crime, but also shows that informal social controls, as reflected by collective efficacy, are key for understanding crime on high crime streets. We argue that it is time for police, other city agencies, and NGOs to begin to work together to consider how informal social controls can be used to reduce crime at residential crime hot spots.