Food insecurity remains a severe threat for people living in least developed and conflict-affected settings. Development organizations frequently apply small-scale agricultural support to strengthen food security in such settings. However, impact evaluation is challenging in these fragile contexts. Thus, rigorous evidence remains scarce and inconclusive. Applying a difference-in-differences approach together with propensity score matching on household survey data from South Sudan, I test the impact of small-scale agricultural interventions coupled with a cash transfer on food security outcomes in a least developed setting highly affected by conflict. My findings emphasize that after one year, the treatment significantly improved household food security measured by the Minimum Dietary Diversity Score. On average, treatment households consumed one additional food type per day due to the support. However, in absence of the cash component, the agricultural intervention did not improve dietary diversity substantially. Other food security measures do not show significant impacts of the combined intervention. The results indicate that the support was effective in maintaining a constant harvest output while control households faced a clear declining trend. Covariate adverse shocks like floods and droughts do not show heterogeneous treatment effects while idiosyncratic shocks such as a disease of the household head inhibit the positive programme impact on dietary diversity. The results imply that small-scale agricultural interventions contribute to diversify household diets in least developed and conflict-affected settings in the short-run.