Over the past several decades minimum wages have steadily gained importance. In many cases this reflects the weakness of unions which have been unable to prevent very low wages compared to the national average wage in some segments of the labour market. Changes in minimum wages can affect employment, income distribution and price level. Empirical investigations in a large number of countries and historical periods show that there is no clear relationship between minimum wages and unemployment. However, there is a broad consensus that minimum wages change income distribution in favour of low-paid workers. Price level effects of minimum wages have not been in the centre of the empirical research. In principle, for the neoclassical paradigm minimum wages have negative employment effects. These iron law of neoclassical thinking came under discussion after negative employment effects of minimum wage increases in empirical studies were difficult to find. The monopsony case was a way out of the dilemma. However, it seems to lack sufficient relevance to draw general macroeconomic conclusions. In the Keynesian paradigm nominal wages become the nominal anchor for the price level. Minimum wages compress the wage structure and lead to a change in income distribution first of all within wage earners. Minimum wage policy should be in line with the following principles: a) Minimum wages must affect a sufficient number of employees, b) they should be adjusted frequently, c) they should increase at least according to trend productivity growth plus the target inflation rate of the central bank, d) they should increase at least in line with average wages because this is the only possibility to prevent an increase in the wage gap. As long as low wages are considered to be too low in comparison to average wages, minimum wages should increase faster than average wages. In a Keynesian perspective no relevant positive or negative employment effects resulting from changes in minimum wages can be expected.