Intercropping is a common practice in traditionally low productivity smallholder farming across Africa. However, as technological improvements and farmers’ investments increase the level of crop productivity, whether intercropping or sole cropping systems remain the best option, is not known. This study researched this question by using empirical experimentation and modelling techniques in two contrasting agro-ecologies in Ethiopia (a humid and a semi-dry environment). The focus of the work was on: (i) understanding farmers’ drivers for the adoption of intercrops or sole crops; (ii) empirically quantifying the efficiency in the use of production resources in sole cropping and intercropping; and (iii) extrapolating the new insights using a cropping systems model, across a range of future (expected) climates for two agro-ecological regions of Ethiopia. A household survey was conducted across two contrasting agro-ecologies (semi-arid and sub-humid) in Ethiopia, to understand farmers’ motives for the adoption of intercropping or sole cropping. The results of the household survey analysis showed that resource poor farmers tended to practice more intercropping than resource rich farmers. In addition, farmers from low yield potential environments were more likely prefer intercropping than farmers in high yield potential environments. The results from a Tobit regression model showed that households that had less labour and a smaller farm sizes were more likely to practice intercropping in both the semi-arid and sub-humid environments. This confirmed that the adoption of intercropping is related to both socio-economic and agro-ecological conditions. Field experiments were carried out in Australia to quantify the efficiency in the use of resources of three cropping systems, these being maize intercropped with navy bean, sole maize, and sole navy beans, under low, medium and high levels of resource availability. The different levels of resource availability were created by combining levels of N and water supply, over two seasons of experiments. The results showed that grain yields of maize and navy bean crops increased with increased levels of resource inputs, irrespective of the crop configuration. However, the performance of the intercropping system, measured by the land equivalent ratio (LER), and indices of resource capture and use efficiency, was generally higher under the low than high levels of resources inputs. This result was in agreement with the survey results that indicated that the advantage of the intercropping systems tended to decrease with increased level of resource availability (e.g. crop inputs). Resource capture (water, nitrogen and solar radiation) and use efficiency tended to decrease in the intercropping system with increased levels of resource availability. Nitrogen capture and water use efficiency were significantly and positively related to the values of LER. As for grain yield, LER values for protein showed that intercropping was more efficient under lower than higher levels of productivity. APSIM explained 85% and 84%, respectively, of the variability observed in maize (n=50), and navy bean yields (n=50). The model results showed that sole bean crops achieved higher grain yields in the semi-arid than in the sub-humid regions, while the yield of the intercropping system and the sole maize system were much higher in the sub-humid environments. Season to season yield variability showed that under sole cropping the probability of not achieving maize yields to satisfy household food security needs (∼500 Kg/ha) was around 15% in a typical semi-arid region under unfertilized soil condition. In sub-humid region, the simulated maize yields could satisfy the food requirement for family household at all levels of N availability. In addition, the simulated advantage in intercropping, in terms of LER, appeared to be greater at lower and medium, than at higher levels of environmental productivity. Simulated results showed that for both agro-ecologies, the intercropping system produced larger total calorie levels than both sole crops. In terms of total protein production and cash returns, the intercropping system was more advantageous over the sole crops in the sub-humid region. In the semi-arid region, the sole bean was advantageous over the intercropping system and sole maize. The likely impacts from expected changes in climate were simulated by running a sensitivity analysis on changes in rainfall and temperature records. Using the period 1982 to 2011 as a baseline, changes in daily rainfall by -10%, -20% and -30% had little or no effect on both maize and bean crops. However changes in average daily temperature by 1oC, 2oC and 3oC caused significant yield reductions in both crops and in both regions. The intercropping system and sole maize cropping appeared to be more resilient as average temperatures increased (up to 3oC) from the baseline. In conclusion, intercropping systems are more resource efficient and better suited for low productivity environments. As the level of farm investment and crop productivity increases, agricultural development projects and extension services need to consider whether sole cropping should replace prevailing intercropping systems in high potential productivity environments (e.g. sub-humid regions).