This article tests the Feldstein-Hoiroka 'puzzle' for a sample of 49 countries from two angles; that of co-movements in the saving-investment series and the magnitude of the saving-retention coefficient (SRC). The overall conclusion we obtain from the results of three different cointegration tests leaves no room for 'puzzle' in the majority of the countries included in the sample. On the other hand, various statistical tests applied to the SRCs show that there are significant differences among the high-income, middle-income and low-income country groups with regards to the salient aspects of capital mobility. First, there are significant differences in the average SRCs of the country groups. Second, the SRCs are differently distributed around their means among the country groups. The results yielded by a third test dictate that future research should pay more attention to the underlying distribution of the SRCs within country groups when testing the 'country size' argument.