The dominant assumption in economic models of climate policy remains that adaptation will be implemented in an optimal manner. There are, however, several reasons why optimal levels of adaptation may not be attainable. This paper investigates the effects of suboptimal levels of adaptation, i.e. adaptation restrictions, on the composition and level of climate change costs and on welfare. Several adaptation restrictions are identified and then simulated in a revised DICE model, extended with adaptation (AD-DICE). We find that especially substantial over-investment in adaptation can be very harmful due to sharply increasing marginal adaptation costs. Furthermore the potential of mitigation to offset suboptimal adaptation is investigated. When adaptation is not possible at extreme levels of climate change, it is cost-effective to use more stringent mitigation policies in order to keep climate change limited, thereby making adaptation possible. Furthermore not adjusting the optimal level of mitigation to these adaptation restrictions may double the costs of adaptation restrictions, and thus in general it is very harmful to ignore existing restrictions on adaptation when devising (efficient) climate policies.