We tested an architect with a lesion to the right prefrontal cortex in a real-world architectural design/planning task that required him to develop a new design for our lab space and compared his performance to an age- and education-matched architect. The patient understood the task and even observed that "this is a very simple problem." His sophisticated architectural knowledge base was still intact and he used it quite skilfully during the problem structuring phase. However, the patient's problem-solving behaviour differed from the control's behaviour in the following ways: (1) he was unable to make the transition from problem structuring to problem solving; (2) as a result preliminary design did not start until two thirds of the way into the session; (3) the preliminary design phase was minimal and erratic, consisting of three independently generated fragments; (4) there was no progression or lateral development of these fragments; (5) there was no carry-over of abstract information into the preliminary design or later phases, and (6) the patient did not make it to the detailing phase. This suggests that the key to understanding our patient's deficit is to understand the cognitive processes and mechanisms involved in the preliminary design phase. We appeal to a theory of design problem solving (Goel, 1995) that associates cognitive processes involved in preliminary design with "lateral" state transformations and argues that "ill-structured" representational and computational systems are necessary to support these transformations. We conclude that the neural basis of this system is selectively damaged in our patient.