Ethnic diversity is portrayed in the literature as a threat to a community’s ability to regulate the behaviour of its members. While there is no shortage of studies examining the effects of ethnic diversity on the social processes important for crime control, findings are inconclusive across national contexts. Further, definitional issues associated with ‘ethnicity’ make cross-cultural comparisons difficult. Using Australian Community Capacity Study survey data from 4091 respondents in 147 Brisbane suburbs, combined with census and police incident data, multivariate regression techniques are utilised to determine the extent to which ethnic diversity influences collective efficacy once we control for other known correlates; and which aspect of diversity ‘matters most’ to levels of collective efficacy. Specifically, we consider the relationship between the diversity or concentration of language, religion and country of birth and collective efficacy. Results indicate that the presence of language diversity and indigeneity in the community are most detrimental to collective efficacy.