This paper examines the relationship between the occurrence of crime on residential areas, namely street robbery and residential burglary, perception of security and streets segments attributes such as connectivity, length, street lighting, appearance, escape routes, presence of security guards, residences visual and physical barriers, visual and physical connections between residences and the public open space, land use and type of housing unit, in the residential boroughs of Menino Deus and Rio Branco in Porto Alegre. Crime has been related not only to socio-economic or political variables but also to physical or spatial variables in distinct urban environments. Some studies have shown that urban configuration tends to play a role in increasing or decreasing opportunities for crime in distinct cities. However, it is necessary to better understand their effect on street robbery and residential burglary, specifically, in the context of residential boroughs in a Brazilian city, and considering a high resolution analysis on the level of a street segment. This type of analysis has been developed and initially used by Hillier & Sahbaz in their paper entitled ‘High Resolution Analysis of Crime Patterns in Urban Street Networks: an initial statistical sketch from an ongoing study of a London borough’, where they emphasize the importance of high resolution analysis of crime patterns in urban street networks to which many physical variables are related. Data related to occurrence of street robbery and residential burglary were collected in the Public Security Department of State of Rio Grande do Sul, for a period of ten months and through questionnaires applied to a sample of 123 residents in Menino Deus Borough and 86 in Rio Branco Borough. This method also allowed the gathering of information concerning residents’ satisfaction levels with security. Street robbery was measured considering the ‘time risk’ which takes into account the length of time a moving person spends on a segment, this being, a function of its length. Residential burglary was measured considering the ‘true rate for the risk bands’, that is, the total number of burglaries over the total number of residential units for each band, each being made up by segments having a common number of dwellings. Land use, identified as residential or non residential, and types of housing units, if houses or flats, were obtained in the Local Department of Planning. Segments characteristics such as street lighting, appearance, presence of security guards, visual and physical barriers, visual and physical connections between residences and the public open space, were obtained trough physical measurements carried out in some segments with higher and lower crime rates. Data was registered in ArcGIS, allowing the generation of maps and the consequent visualization, for example, of the spatial distribution of the two types of crimes, as well as in the SPSS/PC software, where data was statistically analyzed according to the street segments. Segments connectivity and length were numerically and visually generated in the Depthmat software, where a segments map was produced from the axial map representing the street network of the two boroughs. Results reveal, for example, the impact made by segments characteristics such as connectivity, land use and type of housing units on occurrence of crime on residential areas, allowing a better understanding of the relationship between street robbery and residential burglary and some physical characteristics of the urban space. Moreover, not only the occurrence of crime but also the perception of crime on residential areas is related to the use and qual- ity of urban space. Therefore, these results help to deepen the understanding of how to better design for urban security in residential areas.