This paper examines whether the welfare gains from technological innovation that reduces future abatement costs are larger or smaller than the "Pigouvian" welfare gains from optimal pollution control. The relative welfare gains from innovation depend on three key factors - the initially optimal level of abatement, the speed at which innovation reduces future abatement costs, and the discount rate. We calculate the welfare gains from innovation under a variety of different scenarios. Mostly they are less than the Pigouvian welfare gains. To be greater, innovation must reduce abatement costs substantially and quickly and the initially optimal abatement level must be fairly modest.