Background: In Australia, training for general practice (GP) occurs within private practices and their involvement in teaching can have significant financial costs. At the same time there are growing demands for clinical places for all disciplines and for GP there is concern that there are insufficient teaching practices to meet the demand at the medical student, prevocational and vocational training levels. One option to address this may be to change how teaching occurs in the practice. A question that arises in posing such an option is whether different models of teaching change the costs for a teaching practice. The aim of this study is to determine the net financial outcome of teaching models in private GP. Methods: Modelling the financial implications for a range of teaching options using a costing framework developed from a survey of teaching practices in South Australia. Each option was compared with the traditional model of teaching where one GP supervisor is singularly responsible for one learner. The main outcome measure was net financial outcome per week. Decisions on the model cost parameters were made by the study's Steering Group which comprised of experienced GP supervisors. Four teaching models are presented. Model 1 investigates the gains from teaching multiple same level learners, Models 2 and 3, the benefits of vertically integrated teaching using different permutations, and Model 4 the concept of a GP teacher who undertakes all the teaching. Results: There was a significant increase in net benefits of Aus$547 per week (95% confidence intervals $459, $668) to the practice when a GP taught two same level learners (Model 1) and when a senior registrar participated in teaching a prevocational doctor (Model 3, Aus$263, 95% confidence intervals $80, $570). For Model 2, a practice could significantly reduce the loss if a registrar was involved in vertically integrated teaching which included the training of a medical student (Aus$551, 95% confidence intervals $419, $718). The GP teacher model resulted in a net remuneration of Aus$207,335 per year, sourced predominantly from the GP teacher activities, with no loss to the practice. Conclusions: Our study costed teaching options that can maximise the financial outcomes from teaching. The inclusion of GP registrars in the teaching model or the supervisor teaching more than one same level learner results in a greater financial benefit. This gain was achieved through a reduction in supervisor teaching time and the sharing of administrative and teaching activities with GP registrars. We also show that a GP teacher who carries a minimal patient load can be a sustainable option for a practice. Further, the costing framework used for the teaching models presented in this study has the ability to be applied to any number of teaching model permutations.