This paper is based on a study of the relationship between physical and social features of the residential environment and threat which we define as residents' perceptions of problems, fear, and crime. Information was collected through a variety of methods, including interviews with 448 residents in 32 blocks in 12 neighborhoods in Baltimore. Subjects were selected in the following way. Our primary sampling unit was the neighborhood, and so, using a listing of existing community organizations and the perceptions of Baltimore City planners and com-munity leaders, we prepared a map of Baltimore neighborhoods. In order to identify high-threat and low-threat neighborhoods we chose a combina-tion of income and home-ownership rate. (Fortunately, as shown later, our findings confirmed that these were appropriate measures.) In this way we identified three categories of neighborhoods: Type l was pre-dominantly lower income and renter. Type 3 was predominantly middle income and owner, and Type 2 was a mixture of Types 1 and 3. (Upper income neighborhoods were excluded from the study.) Sampling with a probability proportional to size, we obtained a sample of twelveneigh-borhoods. We then looked for sample blocks within each of the twelve neighborhoods. We decided on a street-face definition of a block (both sides of the street between cross streets) because a previous study (Brower, 1979) had shown that this was likely to represent a social unit. In order to obtain both high-threat and low-threat blocks, a well-informed resident in each of the sample neighborhoods was asked to nominate blocks of two kinds: those where people worked together and watched out for one another, and those where people just went their own ways. Through this procedure we obtained a list of 104 blocks which we classified as high-threat and low-threat. From these, we chose 32 sample blocks. We listed all the residences in each of the sample blocks and then selected a random sample of 448 households to interview.