Key words: crop production, intensification, extensification, farming systems, tradeoff analysis, maize, legume, manure, fertiliser, southern Africa Soil fertility decline and erratic rainfall are major constraints to crop productivity on smallholder farms in southern Africa. Crop production intensification along with efficient use of chemical fertiliser is required to produce more food per unit area of land, while rebuilding soil fertility. The objective of this thesis was to identify appropriate crop production intensification options that are suitable to the socio-economic and biophysical conditions of selected smallholder maize-based farming systems in southern Africa. Three sites that formed a gradient of intensity of crop and livestock production were selected for the study. Murehwa in Zimbabwe is characterised by the largest intensity followed by Ruaca and lastly Vunduzi both in central Mozambique. In all three sites, maize is a key staple and cash crop.A literature review, field methods based on participatory research, and modelling tools were combined in analysing potential crop production options across an agricultural intensification gradient. A meta-analysis on maize grain yieldunder rain-fed conditions revealed thatconservation agriculture required legume rotations and high nitrogen input use especially in the early years.Reduced tillage without mulch cover leads to lower yields than with conventional agriculture in low rainfall environments. Mulch cover in high rainfall areas leads to smaller yields than conventional tillage due to waterlogging, and improved yields under CA are likely on well drained soils. Crop productivity underconservation agriculture depends on the ability of farmers to achieve correct fertiliser application, timely weeding, and the availability of crop residues for mulching and systematic crop rotations which are currently lacking in southern Africa. Anadditive design of within-row intercropping was compared to a substitutive design with distinct alternating rows of maize and legume (local practice) under no-tillin the Ruaca and Vunduzi communities of central Mozambique. Intercropping increased productivity compared to the corresponding sole crops with land equivalent ratios (LER) of between 1.0 and 2.4. Maize yield loss was only 6-8% in within-row intercropping but 25-50% in the distinct-row option. Relay planting of maize and cowpea intercropping ensured cowpea yield when maize failed thus reduced the negative effects of dry spells. The residual benefits of maize-pigeonpea intercropping were large (5.6 t ha-1) whereas continuous maize (0.7 t ha-1) was severely infested by striga(Striga asiatica). The accumulation of biomass which provided mulch combined with no tillageincreased rainfall infiltration. Intensification through legume intercropping is a feasible option to increase crop productivity and farm income while reducing the risk of crop failureespecially where land limitation. Cattle manure in combination with chemical fertiliser that included N, P, Ca, Zn, Mn were evaluated for their potential to recover degraded soils and to support sustainable high crop productivity in Murehwa, Zimbabwe over nine years. The experiment was established on sandy and clay soils in two field types. Homefields were close to the homestead and relatively more fertile than the outfields due to previous preferential allocation of nutrients. Maize grain yields in sandy soils did not respond to the sole application of fertiliser N (remained less than 1 t ha-1); manure application had immediate and incremental benefits on crop yields in the sandy soils. A combination of 25 t ha-1 manure and 100 kg N gave the largest treatment yield of 9.3 t ha-1 on the homefield clay soils, 6.1 t ha-1 on clay outfield, 7.6 t ha-1 on sandy homefield and 3.4 t ha-1 in the eighth season. Despite the large manure applications of up to 25 t ha-1, crop productivity and soil organic carbon build-up in the outfield sandy soils was small highlighting the difficulty to recover the fertility of degraded soils. Manure can be used more efficiently if targeted to fields closest to homesteads but this exacerbates land degradation in the outfields and increases soil fertility gradients. The NUANCES-FARMSIM model for simulating crop and animal productivity in mixed crop-livestock farming systems was used to perform trade-off analysis with respect to crop residue management, animal and crop productivity in Murehwa, Zimbabwe. Retaining all maize residues in the field led to severe losses in animal productivity but significant gains in crop productivity in the long-term. Yield increased 4 to 5.6 t farm-1 for RG1, and from 2.8 to 3.5 t farm-1 for RG2. Body weight loss was on average 67 kg per animal per year for RG1 and 93 kg per animal per year for RG2. Retention of all crop residues reduced farm income by US37 and US38 per year for RG1 and RG2 respectively.Farmers who own cattle have no scope of retaining crop residues in the field as it results in significant loss of animal productivity. Non-livestock farmers (60% of the farmers) do not face trade-offs in crop residue allocation but have poor productivity compared to livestock owners and have a greater scope of retaining their crop residues if they invest in more labour to keep their residues during the dry season. This study has revealed that crop production intensification options developed without considering the biophysical conditions as well as socio-economic circumstances of farmers are nuisances. External ideas should be used to stimulate local innovations to push the envelope of crop production without creating new constraints on resource use.