This paper studies the effects of the centrality, connectivity and agglomeration of retail establishments on their long-term viability in three cities in the United States, United Kingdom and The Netherlands. As retail is declining in all three markets, there is a dearth of knowledge on the spatial patterns of this decline. This obstructs the substantiation of development decisions and public policy on urban retail retention and growth. Without knowing where stores are most at risk of closing, where can we decide to invest or divest? This paper uses a self-built dataset of store locations and store closures over the span of more than a century in the urban cores of Detroit, Michigan; Birmingham, England; and The Hague, The Netherlands. While taking different paths, all three cities have experienced significant retail decline over the past century. The probability of store closure over time is compared to the metric distance of stores to the retail center of gravity (centrality), store location along well-used streets as measured by their Choice value (connectivity), and the number of surrounding stores (agglomeration). These three comparisons are statistically analyzed using simple line regression, panel regression, and spatial autoregressive probit models. Across these models, store closure is most significantly affected by agglomeration, then by centrality, followed by connectivity. The significance of all three measures is strongest in The Hague, followed by Birmingham and Detroit – two cities that experienced large-scale urban renewal and socio-economic decline.