Normative decision theory offers a wide range of methods to support decisio
ns. Their successful application, however, is confronted with the human dec
ision maker who wants to behave "naturally" while making decisions, even, w
ith support systems. The main goal of this study was to compare unsupported
and supported decision making and to infer requirements for further develo
pment of decision-aiding tools. In an experiment 36 students had to solve a
complex decision task, first without any support and later using the compu
terised decision tool ADELE, a program based on the decision-analytical Mul
ti Attribute Utility Theory (MAUT). Descriptively, various strategies for u
nsupported decision making, e.g. reduction of the problem space and weighti
ng procedures, and their transfer to the supported condition were observed
and analysed. Larger agreement regarding the decision outcome, longer decis
ion time and less subjective certainty with the decision outcome were obser
ved for the computer-supported decision. The results are discussed in light
of consequences for further development of decision-aiding tools and the n
eed for a. combined approach of descriptive and prescriptive decision resea
rch.