Purpose This paper investigates different methodologies of handling co-products in life cycle assessment (LCA) or carbon footprint (CF) studies. Co-product handling can have a significant effect on final LCA/CF results, and although there are guidelines on the preferred order for different methods for handling co-products, no agreed understanding on applicable methods is available. In the present study, the greenhouse gases (GHG) associated with the production of 1 kg of energy-corrected milk (ECM) at farm gate is investigated considering co-product handling. Materials and methods Two different milk production systems were used as case studies in the investigation of the effect of applying different methodologies in coproduct handling: (1) outdoor grazing system in New Zealand and (2) mainly indoor housing system with a pronounced share of concentrate feed in Sweden. Since the cows produce milk, meat (when slaughtered), calves, manure, hides, etc., the environmental burden (here GHG emissions) must be distributed between these outputs (in the present study no emissions are attributed to hides specifically, or to manure which is recycled on-farm). Different methodologically approaches, (1) system expansion (two cases), (2) physical causality allocation, (3) economic allocation, (4) protein allocation and (5) mass allocation, are applied in the study. Results and discussion The results show large differences in the final CF number depending on which methodology has been used for accounting co-products. Most evident is that system expansion gives a lower CF for milk than allocation methods. System expansion resulted in 63- 76% of GHG emissions attributed directly to milk, while allocation resulted in 85-98%. It is stressed that meat is an important by-product from milk production and that milk and beef production is closely interlinked and therefore needs to be considered in an integrated approach. Conclusions To obtain valid LCA/CF numbers for milk, it is crucial to account for by-products. Moreover, if CF numbers for milk need to be compared, the same allocation procedure should be applied. © 2011 Springer-Verlag.