One of the difficulties in environmental design is the lack of direct commu-nication between the designers (experts) and the users (non-experts). Thenon-expert system, obtainable by combining the evalutaion research withthe expert system (type of Al) is proposed in this paper as an aid forbridging the gap between experts and non-experts. An empirical example ofthe model of evaluation for the system is also presented. A major aim of place evaluation research is to provide the knowledge of us-ers' evaluation to designers so that they can explore the optimum solutionfrom experts' point of view to maximize users' satisfaction. To make placeevalution research more valid as an aid for the environmental design, theauthors have developed the two-step research procedure (Saunui and lnui,1984, 1986) which consists of qualitative research methods.The non-expert system proposed In this paper is a product of this two-stepprocedure combined with Al i.e. the expert system. By feeding the physicalconditions of a given envlronmet into this system,, the system predicts us-ers' evalutaion on that environment (overall preference and the evaluationon the major criteria). Therefore,the system enables designers to knowwhether the plan is satisfactory for users, together with its reasons, whichwould be of great help to designers in finding tthe optimum solution in theirdesign.Since the non-expert system should house the statistically condensedknowledge of users' evalutaion to increase relatiabliity, statistical model ofplace evaluation ought to be prepared perior to the construction of actualsystem. An empirical example of the model has been obtained by our two-step procedure on the evaluation of the exterior of multiple housings.The model consists of 8 diagrams representing different types of evaluationobtained by the phenomenolgical sub-grouping approach (Sanui and Inni,1986), and each diagram was produced by the hierarchical regression anal-ysis. Thus, the causal relationship between the overall preference and themajor evaluation criteria as well as their relative weight was obtained bythe multiple regression analysis. The same process was obtained by themultiple regression analysis. The same process was used to establish therelationship between each of these evaluation criteria and the physical con-ditions of the exterior. The reliablity of these diagrams as the evalutionpreduction tool was reasonably high. The multiple correlation coefficientbetween the overall preference scores of the observed and the estimated(from the physical conditions) exceeded 0.8.