Repeat victimization has recently featured prominently in crime reduction research, especially in the UK. Crime prevention efforts which focus on repeats have enjoyed some success. Yet little is known about what generates repeat victimization. While this remains the case, the scope for crime reduction through the prevention of repeated events against the same target will be limited. There is general agreement about the potential relevance of two intertwined theoretical causes, namely event dependence (where the crime event boosts the probability of repetition), and heterogeneity (where the crime event flags an enduring risk, which enduring risk makes repetition likely). This paper attempts to disentangle the effects of event dependence and heterogeneity on repeat personal victimizations across three interviewing periods using the 1994 National Crime Victimization Survey. We employ a (fixed effects random intercept) multilevel model of personal crime counts incorporating individual and household characteristics as well as lifestyle and prior non‐victimization of individuals clustered within households. Our results show that victimization by personal crime, especially over successive time periods, is partly due to measured heterogeneity. Personal crimes experienced by members of the same household are moderately but persistently correlated. This suggests that unmeasured heterogeneity is also implicated in period‐to‐period repetition of personal crime. The combined effects of unmeasured and measured heterogeneity do not exhaust the predictivity of prior for subsequent victimization, meaning that a degree of event dependence is involved.