This paper is a summary of the studies conducted since 2009 focusing on the assessment of ewe welfare with animal based indicators regrouped in four principles: good feeding, good housing, good health and appropriate behavior of species. We tested the feasibility and reliability of each indicator on 10 farms. Then, relevant indicators were measured during the summer and winter periods in 50 farms (one half of the farm ewes spent the winter at pasture and the other half, indoors) to test between herd variability. The inter and intra-observer reliability was good for all indicators related to the first three principles. For the behavioral measures, the inter-observer reliability was good but the intra-observer reliability was low to moderate. The most appropriate indicators to discriminate the farms were the following: in summer, the humidity of the fleece and two parameters of the test of a human and in winter, the humidity of the fleece and the cleanliness of the hindquarters. In winter, we observed in sheepfold, thinner ewes at the end of winter (P<0.01), very dirty ewes at the level of flanks (P=0.048), more dirtier young ewes at the level of the hindquarters (P<0.005), a higher latency of flight during the test of the human (P<0.005) and a higher distance between the herd front and the human in the beginning of winter (P<0.001). On the pasture, more ewes presented a wet fleece (P<0.001). In our study, wintering at pasture did not raise any particular welfare problems compared to sheepfold wintering.