Several years ago Hidalgo and Hernandez reported a curvilinear, U-shaped, relationship between scale of place (apartment, neighborhood, city) and strength of attachment to the place. In this paper four studies are presented, carried out in four Central-European cities, that (1) confirmed the reported curvilinear relationship using five places (apartment, building, neighborhood, city district, city) in three out of four cities and for five items of the Place Attachment Scale, (2) revealed a consistent curvilinear, inverse U-shaped relationship between scale of place and percentage of variance of place attachment predicted by three groups of factors: physical (type of housing, size of building, upkeep and personalization of house precincts, etc.), social (neighborhood ties and sense of security in the residence place), and socio-demographic (age, education, gender, length of residence, family size), and (3) identified strength of direct and indirect effects of the three groups of predictors on attachment to the five types of places. The curvilinear relationship between place scale and place attachment was particularly strong in highly attractive cities and in those scale items that described people's emotional reactions to places whereas a linear relationship was obtained in the least attractive city and in the items that referred to sense of security, amount of control and knowledge of place. In all four cities the best predicted variable was attachment to middle ranges of the place scale (building and neighborhood). The overall best direct predictor of place attachment was neighborhood ties, followed by direct and indirect effects of length of residence, building size, and type of housing. In conclusion it is argued that the usual choice of predictors of place attachment is biased by researchers' interest in the middle scales of place (neighborhood) at the expense of other place scales. In the paper a claim is made that attachments to smaller (apartments, homes) and larger (city) scales of place along with their unique predictors deserve more attention from environmental psychologists.