In order to create a comfortable waiting space, it is important to understand people's preference in the seat-choice behavior. However, the mechanism is complicated so as the seat-choice behavior is largely varying according to many factors such as seat-types, layout of seats, attributes (age / gender) of users, purpose of use, and the occupancy by other users. The purpose of this paper is to investigate what factors affect seat-choice behavior of outpatients in a hospital, and to construct a model to describe their seat-choice behavior.<br><br> First, we performed detailed observation survey on the seat-choice behavior of outpatients in a waiting space of a hospital. Based on the observation results, we identified five factors affecting on outpatients' seat-choice behavior: (1) whether a seat is the intermediate seat of a chair for three persons or not; (2) whether a seat is a sofa-seat on the side with an elbow rest or not; (3) angle among a digital information panel, a table, and a seat; (4) distance from a seat to a destination (consultation room); (5) whether either of adjacent seats is occupied or not. Also, the utility specific to seat-type and their concurrent behavior (such as eating, reading, or writing) were used as explanatory variables in the model. Using these factors, we constructed a seat-choice behavior model based on a Nested Logit Model to describe outpatients' seat-choice behavior in the waiting space.<br> Unknown parameters of the model were estimated using the observation data. While it was difficult to accurately describe individual seat-choice behavior, the type of seat or a seating zone composed of some seats could be estimated precisely. From the estimated parameters of the model, we discussed the influence of principal factors on the seat-choice behavior. For instance, seats located far from the outpatient' destination or near to the seats occupied by other outpatients were less likely to be selected. Also, the utility of sitting on the intermediate seat of a chair for three persons was seven times for men (but three times for women) as much as that of sitting on the corner seats.<br><br> Next, we performed a simulation using the proposed model to predict the seat-choice behavior of outpatients, and we demonstrated the possibility and capability of the proposed model as a supporting system for actual planning of waiting space. Specifically, we demonstrated the effects of guiding the outpatients' seat-choice by setting some priority seats according to the direction of destination. Finally, we proposed the seat layouts by modifying the present layout, and compared them from the viewpoint of the cumulative value and variance of the utility which outpatients obtained. Simulation results showed that the utility of the individual outpatient was almost equally distributed, also the mean value of utility increased the present layout.