After their service life is over, nuclear power plants must be decommissioned. Accordingly, Sweden has a system with segregated funds to cover the costs. Payments to the funds are dictated by the results of recurrent cost estimates. Recently, differences have been observed between different estimations for the Barsebäck BWR:s. Therefore, the Swedish Radiation Safety Authority, who oversees the system, has commissioned the present study with the objective to investigate the reasons. The work comprises analyses of generic deviances as well as specific ones. It was found that the variations are within the ranges observed elsewhere, but that the precision in comparisons between different reactors can be improved. No new cost raisers were identified for the Barsebäck reactors. It was found that the cost estimation community strongly recommends the parametric method for early estimates whilst the cost calculations on decommissioning are mostly based on the bottom-up method. It is proposed that the parametric method be attempted for comparison between different reactors