Smallholder crop–livestock farming systems have an important role to play for food security in Sub-Saharan Africa, but they have to cope with the effects of climate variability and change. In this study, we test the impacts of different interventions in two contrasting mixed farms in Northern Burkina Faso against the background of plausible current and future climate scenarios. For this purpose, we developed a dynamic farm-household modeling framework around existing tools: crop and animal production models APSIM and LivSim, the household model IAT and the climate generator Marksim. The two farms (a small and a larger) were selected and parameterized based on information collected in a household survey. Tested interventions included different crop fertilization and animal supplementation levels, mulching with crop residues and an alternative livestock feeding strategy. Baseline (2013) and a 2050 projection based on IPCC RCP 8.5 describe two climate scenarios (90 years) for comparison. The maximum level of inputs increases farm energy production by + 90% and + 76% compared to the baseline for the small and the larger farm, respectively. Input levels maximizing net incomes are moderate, though higher than those currently used in both farms. The inter-annual distributions of net income show that the use of external inputs increases both upside and downside risks, i.e. the probability of getting both very high and very low results. This is because the interventions are more effective at increasing the highest yields in good years than at preventing the low production levels of some years. The 2050 climate scenario has a negative impact on energy production and potential income, especially for the scenarios with high input levels. Downside risks could partly explain why farmers do not currently use optimal input levels, and the results suggest that these constraints could intensify with climate change.