Until now, two main sets of arguments have dominated the debate on the nature of the massacres that were perpetrated in Rwanda before the 1994 genocide. The first one maintains that they constituted a response to prior attacks by the RPF, implying that they should be regarded as military operations, rather than as acts of ethnic cleansing. The second common line of argument is that these massacres served as pilot runs for the subsequent genocide, implying that they were part of a plan that was not to see its full implementation until 1994. This paper puts forth a third, alternative interpretation of these massacres. The first of the aforementioned arguments, it is argued, does not take into account the detailed evidence that is available on the killings: the fact that they took place in the context of the civil war accounts for the timing of the massacres, but not for their genocidal character. In turn, the second interpretation fails to situate these massacres against the agro-pastoral and ideological background of the regime that committed them. By contrast, this paper shows that the massacres took place in areas characterized by a specific history of spatial and social engineering. They are best understood against the background of the processes of land colonisation, re-settlement, depredation and dispossession of cattle and land that were under way in the areas where the land was most scarce, and where the peasant society was being subject to rationalisation and re- modelling from above. The paper concludes that pastoralism was sentenced to disappear from Rwanda and that the massacres should be considered instances of ethnic cleansing.