The top portions of the root system of deeply rooted plants are frequently in dry soil while deeper roots still have access to water. We expected that many surface roots would be shed when subject to localized soil drying. We further hypothesized that the cost of fine root construction per unit root length would be negatively correlated with the rate at which root length is shed. Seedlings of four citrus root- stocks that varied widely in specific root length (cm g−1 root) were used to test these hypotheses. Plants were grown for 4 months in a split-pot system divided into a top and bottom pot. After roots were well established in the bottom pot, water was withheld from the top pots of half of the plants; plants were harvested every 2 weeks thereafter. Sufficient water was supplied to the bottom pot to prevent shoots of droughted seedlings from experiencing significant water stress. All plants were labelled with 14CO2 48 h before harvesting, and autoradiographs made of the fine roots harvested from the droughted compartment. Comparisons of the autoradiographs with digitized images of the root system allowed us to assess root mortality and root sink activity. As expected, the proportion of 14C-labelled photosyrithate allocated to fine roots in the top pot declined with soil drying in all four genotypes; however, there was no genotypic effect on this decline. Contrary to our expectations, extensive root mortality was not apparent for any genotype, even after 60 d of localized soil drying. Apparently, selection for rapid shedding of roots in response to soil drying has not occurred in these Citrus species.