Unless an active environmental policy exists, firms have no incentive to engage in abatement or environmental R&D so policy design is of paramount importance. This design heavily depends on the way R&D spillovers operate. There are two distinct types of R&D spillover: output spillover and input spillover. An input spillover operates on the expenditure toward pollution reduction, whereas an output spillover manifests as the achieved abatement. Under optimal emissions taxation, significant differences arise due to this distinction, in particular, when the spillover operates on R&D inputs. In an oligopolistic setting, the result is higher R&D expenditure, but also higher aggregate emissions and, consequently, higher emissions taxes. By contrast, when spillovers occur in R&D output, there is a U‐shaped relationship between the optimal tax and the spillover, showing a trade‐off between the optimal tax rate and spillovers when these are low. In terms of the relative effectiveness of different R&D organization setups, combining emissions taxes with R&D cooperation, this paper shows that under low levels of R&D spillover R&D cooperation gives higher emissions reductions, whereas when spillovers are high this is not the case.