For osmotic adjustment to be used as a selection criterion for adaptation to drought there is a need to demonstrate its segregation and association with seed yield in a population. This study examined osmotic adjustment and seed yield in F3 lines derived from crosses between Brassica napus (cvv. Karoo, Monty) and B. juncea (line JN25). A clear separation of the level of osmotic adjustment was observed among 20 F3 lines of Karoo x JN25 (7 high and 13 low) and 20 F3 lines of Monty x JN25 (8 high and 12 low) under glasshouse conditions. When the 3 parents and 8 selected F3 lines were grown in a low-rainfall environment in the field, the segregation of osmotic adjustment was largely similar to that observed in the glasshouse. Yield reduction, with irrigated plots as controls, was up to 30% for genotypes with low osmotic adjustment but only 10% for those with high osmotic adjustment. Osmotic adjustment was closely correlated with the accumulation of K+ (r = 0.91), soluble sugars (r = 0.90) and proline (r = 0.96), whereas other solutes (Na+, NH4+, Cl-, NO3-, Mg-2(+) and Ca-2(+)) made little or no contribution to osmotic adjustment. Proline concentrations were very low in well-watered plants but sharply increased by 5- to 15-fold in plants subjected to water deficit, and net proline accumulation showed a higher association with total seed number (r = 0.86) and yield (r = 0.89) than did the concentrations of K+ (r = 0.75 to 0.82) or soluble sugars (r = 0.68 to 0.72). The study suggests that leaf proline concentration could be a good indicator of osmotic adjustment in Brassica oilseeds.