The concept of collective efficacy, defined as the combination of mutual trust and willingness to act for the common good, has received widespread attention in the field of criminology. Collective efficacy is linked to, among other outcomes, violent crime, disorder, and fear of crime. The concept has been applied to geographical units ranging from below one hundred up to several thousand residents on average. In this paper key informant- and focus group interview transcripts from four Swedish neighborhoods are examined to explore whether different sizes of geographical units of analysis are equally important for collective efficacy. The four studied neighborhoods are divided into micro-neighborhoods (N=12) and micro-places (N=59) for analysis. The results show that neighborhoods appear to be too large to capture the social mechanism of collective efficacy which rather takes place at smaller units of geography. The findings are compared to survey responses on collective efficacy (N=597) which yield an indication in the same direction through comparison of ICC-values and AIC model fit employing unconditional two-level models in HLM 6.