Climate change, increased climate variability and poor soil fertility are major bio-physical constraints to cropping on smallholder farms in southern Africa. We used the Agricultural Production Systems sIMulator (APSIM) to assess maize yield response to integrated soil fertility management (ISFM) rotational sequences of low-quality organic resources, nitrogen-fixing green manure and grain legumes, and mineral fertilizers under baseline (1960–2005) and projected (2040–2069) climates. APSIM was also employed to evaluate the response of maize yield to the ISFM sequences under a combination of a changing climate, and typical sowing dates and mineral fertilizer application rates for smallholder households of varying resource endowment. The ISFM sequences were ‘Fertilizer-start’ [a sunnhemp (Crotalaria juncea (L.)-based sequence], ‘Litter-start’ (a woodland litter-based sequence), and ‘Soya-start’ and ‘Manure-start’ (cattle manure-based sequences). The simulated maize grain yields were used to analyze agronomic and economic risks of maize productivity. Agronomic risk was evaluated on the basis of sufficiency of the maize grain yield to meet annual household calorie (energy) requirements, while economic risk was assessed using gross margins. For model evaluation, the simulated maize yields compared well with those measured from the field experiment (RMSE = 0.11–0.55; R2 = 0.55–0.93). Under the baseline climate, ‘Soya-start’ was the least risky ISFM option as only one (2.2%) of the 45 years had calories lower than the minimum acceptable limit of 4,872,750 kcal required to meet household food self-sufficiency for a family of six people. Conversely, continuous fertilized maize and ‘Litter-start’ were the most risky options among the fertilized treatments as three (6.7%) and four (8.9%) of the 45 years, respectively, yielded calories that were below the threshold. Across treatments, the number of years with maize grain yield exceeding 2.3 t ha−1 was higher under baseline compared with future climate. However, ‘Soya-start’ and ‘Manure-start’ were consistently the least risky options under the future climate. The cattle manure-based sequences also had the lowest economic risk under both baseline and future climates. Over the 45-year period under baseline climate, ‘Soya-start’ only had two years with negative gross margins compared with six and nine for ‘Litter-start’ and continuous fertilized maize, respectively. A similar trend was observed under future climate. Overall, agronomic risk was lowest under sowing dates and mineral fertilizer application rates for resource-endowed (RG1) farmers compared with their resource-constrained (RG3) counterparts. Agronomic risk was higher under the representative concentration pathway (RCP) 8.5 compared with RCP4.5, with an average increase of 6, 4 and 6% for RG1, resource-intermediate (RG2) and RG3 management, respectively. Under the typical farmer management practices, the cattle manure-based sequences gave lower agronomic risk than continuous fertilized maize. Economic risk under the farmer management scenarios overly mirrored trends observed for agronomic risk. We conclude that sequenced ISFM combinations of organic resources, nitrogen-fixing green manure and grain legumes, and mineral fertilizers reduce climate risk in smallholder rainfed crop production systems in southern Africa, and similar agroecologies.