Whereas one line of recent neighborhood research has placed an emphasis on zooming into smaller units of analysis such as street blocks, another line of research has suggested that even the meso‐area of neighborhoods is too narrow and that the area surrounding the neighborhood is also important. Thus, there is a need to examine the scale at which the social ecology impacts crime. We use data from seven cities from around the year 2000 to test our research questions using multilevel negative binomial regression models (N = 73,010 blocks and 8,231 block groups). Our results suggest that although many neighborhood factors seem to operate on the microscale of blocks, others seem to have a much broader impact. In addition, we find that racially and ethnically homogenous blocks within heterogeneous block groups have the most crime. Our findings also show the strongest results for a multitude of land‐use measures and that these measures sharpen some of the associations from social characteristics. Thus, we find that accounting for multiple scales simultaneously is important in ecological studies of crime.