Dairy cattle in New Zealand account for about 17% of the country’s total greenhouse gas (GHG) load. As the nation progresses in its efforts to reduce these emissions, is likely that the dairy industry will face physical constraints and/or financial costs associated with their mitigation. For this study, a non-linear optimization model was developed to analyse the cost-effectiveness of diverse mitigation strategies for reducing GHG loads between 10–30% within systems of different production intensity in the main dairy farming regions of New Zealand: Waikato and Canterbury. Pastoral dairy farms, as found in New Zealand, are complex farming systems with many interdependent management variables and any change in one of these variables in order to reduce GHG emissions will inevitably affect the system as a whole, generating changes in its physical and economic outputs. Computer modelling is thus highly suited to evaluate alternative strategies, given its capacity to represent and consider these interdependencies simultaneously. De-intensification, by reducing any combination of stocking rate, nitrogen fertiliser and supplement use, was a key strategy in reducing GHG emissions in an economic manner across the whole range of farming systems and regions considered in this study, as well as for all levels of emissions constraints tested. The mean farm profit reduction found under this level of mitigation across all studied systems was 6%, when no other mitigation strategies were available. To achieve this, the mean reduction in stocking rate was 8.8% and the mean nitrogen fertilizer reduction was 36%. Nitrogen fertiliser usage increases GHG emissions by increasing the available mineral nitrogen in soil and thus increasing denitrification and also by increasing enteric methane emissions as more fibrous feed is available to be consumed. The mean profit reduction was 6.9% and 4.9% for the Waikato and Canterbury regions, respectively. The Canterbury medium-input system had a lower cost of mitigation than the other systems, associated with the use of more supplements with high embedded GHG emissions in the baseline. Accordingly, the use of this kind of supplement was quickly reduced by the optimization model in the mitigation scenarios. At the other end, the Canterbury high input system had the highest cost of mitigation, arising from a combination of larger GHG-e reductions required in absolute terms and the low profitability of the baseline plan, given its high use of imported supplement. In contrast, the abatement cost in the Waikato systems was intermediate between these two extremes. Improved reproductive performance and improved genetic merit of the herd were selected as optimal in all cases where specific mitigation strategies were available. Profit was reduced by 1.6%, on average, in this case when 10% GHG emissions reductions were imposed. To adapt to this constraint, stocking rate was reduced by 8.1%, while N fertilizer was reduced by 30.7%, on average. Thus, although the availability of specific mitigation strategies was valuable in reducing the profit losses arising from mitigation, these were not sufficient to reduce emissions without some degree of de-intensification. Other mitigation strategies, like removing cows from pasture or the use of denitrification inhibitors, were not always cost effective and only entered the optimal solution when higher levels of mitigation were imposed. This demonstrates that the high cost of adopting these specific mitigation strategies is only warranted when systems are exposed to larger profit reductions associated with higher levels of de-intensification. A quick calculation using average reduction in milksolids production found in this study shows that the loss of revenue of the dairy industry if 30% mitigation of GHG is imposed can represent as much as half the export earnings of the entire beef industry of the nation.