Using a computer simulation, one can determine what the optimum table arrangement would be for restaurants of various sizes that accept walk-in customers only and take no reservations. At issue is whether the restaurateur can gain more revenue when its tables are dedicated to seating parties of specific sizes (for example, parties of one and two people would be served at 2-tops, while parties of one to four people would be served at 4-tops) or whether the restaurant should use tables that can be combined as needed according to party size. The simulation predicted that combinable tables would prove most useful in a small restaurant with a small average party size. Combining tables in that situation increased revenue per available seat hour by about 2 percent compared to having only dedicated tables. In a large restaurant or any restaurant with a large average party size, the simulation found that dedicated tables were superior to combinable tables. A loss in productivity occurs when some number of tables are held out of service until adjacent tables become available (so that the tables can be combined to seat a large party). The simulation found that the most efficient approach is for a restaurant’s table-size mix to match its customer party size mix, since doing so increases the restaurant’s effective customer-service capacity. However, that customer mix cannot always be known before a restaurant is constructed, and that mix might change during different dayparts. Moreover, the simulation makes certain assumptions that may need further examination, and it does not take into account such aesthetic factors as customers’ reactions to a particular restaurant layout.