A recent study of the livestock water productivity (LWP), at higher spatial scales in the Blue Nile Basin, indicated strong variability across regions. To get an insight into the causes of this variability, we examined the effect of farm households’ access to productive resources (e.g. land, livestock) on LWP in potato–barley, barley–wheat, teff–millet and rice farming systems of the Gumera watershed (in the Blue Nile Basin, Ethiopia). We randomly selected 180 farm households. The sizes of the samples, in each system, were proportional to the respective system’s area. Then we grouped the samples, using a participatory wealth ranking method, into three wealth groups (rich, medium and poor) and used structured and pre-tested questionnaires to collect data on crops and livestock management and applied reference evapotranspiration (ET₀) and crop coefficient (K_c) approaches to estimate depleted (evapotranspiration) water in producing animal feed and food crops. Then, we estimated LWP as a ratio of livestock’s beneficial outputs to water depleted. Our results suggest strong variability of LWP across the different systems: ranging between 0.3 and 0.6 US$ m^-3 year^-1. The tendency across different farming systems was comparable with results from previous studies at higher spatial scales. The range among different wealth groups was wider (0.1 to 0.6 US$ m^-3 year^-1) than among the farming systems. This implies that aggregating water productivity (to a system scale) masks hotspots and bright spots. Our result also revealed a positive trend between water productivity (LWP and crop water productivity, CWP) and farm households’ access to resources. Thus, we discuss our findings in relation to poverty alleviation and integrated land and water management to combat unsustainable water management practices in the Blue Nile Basin.