Irrigation treatments were imposed in the field on an indeterminate cultivar of narrow-leafed lupins (Lupinus angustifolius L., cv. Danja) and on a breeding line with reduced branching (75A/329) so that they experienced no water-deficits (frequently irrigated), a transient mild water-deficit or a transient severe water-deficit during early reproductive growth, or continuous severe water-deficit during reproductive growth (unirrigated). Both leaf water potential and leaf conductance declined in all treatments in which a water-deficit was imposed. Differences in leaf conductance were apparent before differences in leaf water potential: Conductance declined to 40% and 30% of the frequently irrigated controls in the transient mild and severe water-deficit treatments, respectively. Leaf water potential declined to -1 - 1 MPa and -1.6 MPa, respectively, in the transient mild and severe water-deficit treatments, compared to between -0 - 65 and -0 - 95 MPa for the frequently irrigated controls. Seed yield and total dry weight were reduced in the transient severe water-deficit and unirrigated treatments, but were no different from the frequently irrigated treatment when the water-deficit was transient and mild. However both transient water-deficit treatments produced more main-stem seed yield than the frequently irrigated treatment, especially in the reduced-branching line 75A/329. The transient mild water-deficit treatment also produced more first-order apical axis yield than the frequently irrigated treatment. These yield increases were mainly due to a greater yield of seed per pod, although on the first-order apical axes there was also a tendency to set more pods. The greater seed yield per pod in the transient water-deficit treatments was due to an apparent redirection of assimilate from vegetative to reproductive growth. This was not due to a smaller reduction in reproductive growth rates than in vegetative growth rates, but to an acceleration of reproductive growth that was maintained after stress relief. The same early acceleration of reproductive growth was also observed in unirrigated treatments, but the severe stress which persisted throughout later reproductive growth reduced pod growth rates and negated the early advantage.