Cowpea production is severely affected by environmental stress factors, particularly drought. More drought-tolerant crops are therefore urgently required for future improvement of food production in Mozambique. To increase high productivity and sustainability of cowpea there is a need to establish an active local breeding program, this should also include screening and characterisation of germplasm to select for more drought-tolerant cowpea landraces. This study has been therefore conducted in a temperature-controlled greenhouse at the University of Pretoria and in an open-sided tunnel house in Mozambique. The overall aim of the research work was to identify the most drought-tolerant cowpea landrace currently deposited in the Mozambican gene bank. Results of this study showed that tested Mozambican cowpea landraces have a different degree of drought tolerance with one Mozambican cowpea landrace, Timbawene moteado, better performing under drought conditions. In particular, plants of this landrace had more vigorous growth better overcoming a drought period with fast recovery and re-growth after drought exposure. Plants also maintained high rates of photosynthetic CO2 assimilation when exposed to drought and used better assimilated carbon to generate biomass than the other tested cowpea landraces. Protein and chlorophyll degradation was further less affected and had only a slight increase in proteolytic activity under drought with the proline content significantly increasing under drought. In contrast, plants of the landrace Massava nhassenje were the most drought-sensitive plants with low water-use efficiency and low CO2 assimilation as well as the lowest shoot biomass accumulation and high protease activity under drought. This study has overall demonstrated that the Mozambican cowpea germplasm deposited in the seed bank is diverse and contains characteristics that could be useful for a national breeding program. Shoot biomass, were thereby valuable traits which could be easily measured in Mozambique in a tunnel house experiment. This study might serve as a basis to screen a greater number of landraces to identify a greater number of landraces as useful additions in an active local cowpea breeding program.