Three different drought stress levels (water potential of the nutrient solution, Ψ s = −0.6, −1.2 and −1.8 MPa, respectively), and a control treatment ( Ψ s = −0.1 MPa), were applied during 2 weeks to three almond species, followed by 3 weeks of recovery. The selected test species were Prunus dulcis (Miller) D.Webb (bitter almond) and two wild almond species, P. lycioides (Spach) C.K. Schneider and P. scoparia (Spach) C.K. Schneider. All three are species native to Iran, and can be used as rootstock, but only P. dulcis is actually used for commercial almond production. In the absence of drought stress, maximum net assimilation rate ( A max) is highest for P. scoparia and lowest for P. dulcis. For all species A max was above 16 μmol CO 2 m −2 s −1. A similar relationship between A max and dark respiration rate ( R d), was observed for all species. This relationship suggests that optimisation of the carbon budget is independent of species. The three investigated species seem to have a different reaction to a similar stress, indicating different drought stress coping strategies. P. scoparia lost all its leaves during the experiment, while P. lycioides only kept some leaves, however, the remaining leaves were almost totally wilted and did not allow for any photosynthesis measurement. P. scoparia did not recover during the experiment, as no new leaves were developed once Ψ s was restored to pre-drought stress levels. However, this species has green stems, indicating that stem photosynthesis might play an important role in the plants’ overall carbon balance. This species is an opportunistic one ( sensu [Higgins, S.S., Larsen, F.E., Bendel, R.B., Radamaker, G.K., Bassman, J.H., Bidlake, W.R., Alwir, A., 1992. Comparative gas-exchange characteristics of potted, glasshouse-grown almond, apple, fig, grape, olive, peach and Asian pear. Sci. Hortic. 52 (4), 313–329]), where assimilation is seriously limited by non-stomatal processes as evidenced by measurements of intercellular CO 2 concentration, eventually resulting in total leaf loss. All P. lycioides leaves almost completely wilted during the experiment, but this species recovered rather quickly. Leaves, newly formed at the end of the experiment, obtained maximal assimilation rates under control Ψ s levels, equivalent to those measured in the control treatment. Finally, P. dulcis did keep at least part of its leaves during drought stress. However, assimilation rates after 2 weeks of drought treatment and 3 weeks of recovery were only about half of those measured in the control treatment. Of the three investigated species, non-stomatal limitation of assimilation seems to be the least important in P. dulcis. Intrinsic water use efficiency, defined as the ratio of assimilation rate over stomatal conductance, increased for P. dulcis with increasing drought stress, while a different pattern was observed for P. lycioides and P. scoparia, indicating non-stomatal processes prevail over stomatal limitations of the assimilation process. It was concluded that P. dulcis is the species most tolerant to drought. P. scoparia tries to avoid drought, whereas P. lycioides has an intermediate behaviour. Besides P. dulcis, also P. lycioides seems to have some potential for use as rootstock for commercial almond production.