The primary purpose of this research was to subject Conlon and Garland's (1993; Garland & Conlon, 1998) project completion hypothesis to a stronger test than it has faced in the past. We conducted an experiment in which we manipulated the degree of completion, sunk-cost amount, and anticipated sales price for a hypothetical real estate development project. The sales price information allowed participants to calculate the economic value of the project. Participants were also held "accountable" in the sense that they had to explain the reasoning behind their decisions to the experimenter. Together, the sales price information and accountability stipulation created an inducement to engage in an economically rational decision process. In spite of this inducement, participants were unduly affected by the project's closeness to completion. In fact, when the project was close to being finished, they often recommended completing the project even when it was economically unwise to do so. While the effects of the project completion manipulation were surprisingly strong, the sunk-cost manipulation had virtually no effect. Implications of these results and directions for future research are discussed. Copyright 2000 Academic Press.