Fuel cells are high efficient energy production systems producing heat and electrical power. Especially PEM fuel cells (Polymeric Electrolyte Membrane) require pure hydrogen for operation. Hydrogen is produced e.g. by reforming of hydrocarbons like natural gas or ethanol. During the production of hydrogen carbon monoxide is formed. CO is a harmful poison for the platinum catalsyst of the fuel cell electrodes. Therefore hydrogen has to be purified. The goal of this work was to develop a pressure swing adsorption for small scale fuel cell systems. The hydrogen to be purified is produced by reforming bio-ethanol. The PSA is designed for a quite low adsorption pressure of 7 bar and a small thermal hydrogen power for fuel cell systems in a range of 1 kW electrical. For the design of the PSA a simplified method of calculating the bed height of the adsorbers was developed. Most scientists working in the field of pressure swing adsorption use mathematical models of breakthrough curves for PSA calculations. But these models can only be solved numerically. The results of these simulations are approved by performing test at lab-scale PSA. Main focus on the tests is set to the breakthrough time and the shape of the breakthrough curves. The calculation method developed in this work is also verified by test. But no breakthrough curves were taken up and no breakthrough time of the unwished gas component (CO) was measured. Only the hydrogen quality and the hydrogen recovery rate even after several adsorption cycles were the dominating criteria for the approval of the calculation model. The PSA tests showed that the required hydrogen quality (less than 10 ppmv CO in hydrogen) for PEM-fuel cells was reached even after a 20 adsorption cycles (one cycle lasted 12 minutes with 3 minutes adsorption). The tests were performed on several days in series without changing the adsorbents or purging them for a longer time.