Two experiments studied the ways dogs perform on concurrent operant schedules when choosing between different amounts of food. Experiment 1 allowed six dogs to choose between obtaining different amounts of food available for responding on two levers. Across phases of the experiment, the food amounts were available in four different weight-to-weight ratios, 1:1, 2:1, 3:1, and 4:1. The operant schedule was concurrent VI 60 sec VI 60 sec. The results showed that the dogs “undermatch” when the ratio of their responses on the two levers is plotted against the ratios of the amounts of food they produced by responding on those levers. When the response ratios were recomputed to include only those lever-presses that occurred outside the change-over delay (C.O.D.) period, the severity of undermatching was reduced, and the data averaged across all six dogs showed a matching function. Experiment 2 studied the effect of varying the duration of the C.O.D. when the dogs chose between food-amounts in a 3:1 weight-to-weight ratio. The number of times the dogs alternated between responding on the two levers per session was inversely related to the duration of the C.O.D., but the approximation to matching seems little affected.