Burglary prevalence within neighbourhoods is well understood but the risk from bordering areas is under-theorised and under-researched. If it were possible to fix a neighbourhood's location but substitute its surrounding areas, one might expect to see some influence on its crime rate. However, by treating surrounding areas as independent observations, ecological studies assume that identical neighbourhoods with markedly different surroundings are equivalent. If not, knowing the impact of different peripheries would have significance for crime prevention, land use planning and other policy domains. This paper tests whether knowledge of the demographic make-up of surrounding areas can improve on the prediction of a neighbourhood's burglary rate based solely on its internal socio-demographics. Results identify significant between-area effects with certain types of periphery exerting stronger influences than others. The advantages and drawbacks of the spatial error and predictor lag model used in the analysis are discussed and areas for further research defined.