Rationalization of social policy by im provement in social benefits targeting is a popular idea in Poland and in other countries. I argue that it is not so simple task as it looks like. First, we have targeting problems in universal social benefits, for instance problem of non take-up. Second, connection between two types of targeting errors has an interesting feature described by K.R. Hammond’s ‘iron law ’ of social policy. We can decrease probability of one of them at a cost of increase in probability of the second one. Another dimension is a hope that im provement in targeting will increase the level of benefits. It may be vein because political support for very selective programs is weak and that fact influences amount of money allocated to them . Finally I joined Hammonds law with Martin Rein’s ‘law ’ (if we have constant amount we can divide it in two ways: more to less people or less to more people) and evaluated in this context Charles Murray’s law s of social welfare.