In West Germany during the 1980's, law changes cut the level of unemployment compensation for the unemployed without children, and extended the duration of unemployment insurance for unemployed aged 42 and over, leaving other groups unaffected. This paper takes advantage of these natural experiments to estimate the disincentive effects of unemployment compensation in Germany, using the German Socio-Economic Panel. The analysis of the cut in level is inconclusive. However, the extension of the duration by as much as three times for some older unemployed had a large impact, decreasing the escape rate from unemployment for affected groups by 44-62 percent. Transitions to out of the labor force and transitions of unemployed under 50 were more responsive to extensions of unemployment insurance than transitions to employment and transitions of unemployed 50 or over. The implied elasticity for escapes of men to employment is similar to estimates for men in the United States. The elasticities imply that differences in the potential duration of unemployment insurance only partially explain the difference in spell lengths between Germany and the United States.