In 1995 the statutory sickness fund (AOK) in Magdeburg arranged target agreements with 10 of 23 acute care hospitals in its district to exercise an influence on the development of the average length of hospital stay. With the aim of decreasing the length of stay as well as the administrative cost and effort, these agreements set upper limits on the average length of stay which were hospital-specific and period-specific. In return, with only a few exceptions, the AOK Magdeburg refrained from limiting the coverage of individual cases. Hospital cases discharged from 1994 and 1996 were analysed to determine whether the development of the length of stay in the ten hospitals with target agreements differed from that in the other 13 hospitals. Only some of the hospitals were successful in reaching their target agreements. The average length of stay dropped by 4.2% in the hospitals with target agreements and by 7.9% in those without target agreements. This must be considered in the context of the development of the case load and number of available hospital beds. For instance, in spite of a target agreement, one hospital showed a (compensatory) increase in the average length of stay in association with an increase in the number of authorized beds and a concurrent decrease in the number of cases. The number of days billed by AOK patients per authorized bed (as an indicator of hospital productivity) showed a more favourable development in the group of hospitals with target agreements than in the other group. This was not a controlled trial as far as the selection of the hospitals is concerned. The results suggest that there is no harm in incentives that induce hospitals to manage primarily on their own the average length of stay. The use of routine aggregate data in monitoring this development, rather than the current more expensive individual case approach, also seems reasonable. Well planned studies that further test the "tool" of target agreements can be recommended.