Grandmothers provide more childcare for their daughters' children than for those of their sons, almost everywhere. Exceptions occur where virilocal (patrilocal) postmarital residence makes the children of sons more accessible, but even under virilocality, preferential care of daughters' children, net of the effects of proximity, is often demonstrable. A unique counter-example has been reported by Pashos (2000; Evolution & Human Behavior, 21, 97–109) who found that rural Greek grandmothers cared more for their sons' children even when effects of proximity were controlled; however, this result was based on an analysis in which everything from living in the same household to living in nearby villages was treated as equally close. Here, we present new analyses that replicate Pashos's result, based on a large European survey with a finer differentiation of residential proximity. In interviews conducted in 2004-2007, rural, but not urban, Greek women indeed reported more care of sons' than of daughters' children, net of the effects of proximity and other variables, This rural reversal of the usual uterine (matrilateral) bias was not observed elsewhere in Europe. Greeks were not surveyed again until 2015, whereupon the pattern had disappeared, with rural women now exhibiting a strong uterine bias in grandchild childcare. It seems likely that the financial crisis of 2008-2009, which hit Greece especially hard, played some role in this dramatic change, but it cannot readily be traced to increases in either unemployment or multigenerational households.