The absence of data at the correct spatial scales has made testing for the neighbourhood effect in voting at British general elections very difficult to undertake. A specially-constructed data set for ‘bespoke neighbourhoods’ at a range of small scales makes such tests feasible for the 1997 general election. These are used in conjunction with individual data obtained from the British Election Study to test whether voting by housing tenure varied according to the composition of the local housing market. The results provide evidence that is entirely consistent with the hypothesised effect (although it is only circumstantial regarding the reason for it), and strongly indicates that how people vote is substantially influenced by the nature of their (very) local milieux.