New forms of state income support for f armer's require knowledge of the statistical distribution of farming households according to their total income. However, France has no reliable means of analysing that distribution y et, particularly of low income groups. Realized improvements are very small as we see in revisiting measurement of farmers' Personal incomes. Households here considered are those offarmers working exclusively or principally in agriculture. Four sources of microdata are examined : the Farm Accountancy Data Network (RICA in France), fax data, household surveys with interview, other opérations such as the '"analysis" of Farming Statistics. RICA is the best tool of agricultural income' s measurement for professional farms. With the pluriannual resuit s it is possible to distinguish transient low incomes and persistent low incomes. So we know that if 35 % of holdings belonging to network from 1982 to 1986 (n = 2207) have a negative or zero net income (after payment of compulsory social insurances contributions) during a year at least, more than half of deficits last only one year. If 66 % of holdings have a small net income, i.e. lower than 40 000 F, during one year at least, only 25 % — but it's still much — have such an income during five years. But the lowest area of distribution is partly excluded, smallest holdings being out of survey field. Another restriction is that RICA in France has not for target to capture family incomes from non agricultural origins. If surveys on fiscal revenues include incomes of ail origins, underestimation is important and varying among households. Number of low total incomes is overestimated. However it is the less imperfect source on farmer' s non agricultural incomes : wages are better known ; and error margin being less for selected categories of households, these results can be useful to support assumptions on amounts per categories of total fiscal and non fiscal revenues. Although for few years household interview surveys have several questions on incomes, their results on this point for farmers are disappointing. "Analysis" of farm statistics is an evaluation of global income (farm income + other gainful activity return + retirement pension) obtained by families living or working in farms, for selected years. Several of family incomes are calculated by standards. So data are potential incomes at level ofeach family ; but groups' means allow to estimate some disparities. As to implementation of agro-social policy, it refers to Personal incomes which usually are roughly estimated. This review leads to specify what partial studies of personal distribution are today possible for farm and non agricultural income offarmers working exclusively or principally in agriculture. Suggestions are made to attempt an improvement of present measurement.