Previous reports indicate that ribulose 1,5-bisphosphate (RuBP) binds very tightly to inactive ribulose bisphosphate carboxylase (rubisco) in vitro. Therefore, we decided to investigate whether there was evidence for tight binding of RuBP associated with deactivation of rubisco in vivo. We modified a technique for rapidly separating ;free' metabolites from those bound to high molecular compounds. Arabidopsis thaliana plants were illuminated at various irradiances before freezing the leaves in liquid N(2) and assaying rubisco activity and RuBP. The percentage activation of rubisco varied from 37% at low irradiance (45 micromoles quanta per square meter per second) to 100% at high irradiance (800 micromoles quanta per square meter per second). The total amount of RuBP did not vary much with irradiance, but bound RuBP changed from 36% of the total at low irradiance to none at high irradiance. Bound RuBP was significantly correlated with the estimated number of inactive rubisco sites, with a ratio of about 1:1. After a step increase in irradiance, rubisco activation increased and total RuBP increased transiently, but steady levels of both occurred by 10 minutes. The amount of bound RuBP decreased with a similar time course to the estimated decrease in inactive rubisco sites. After a step decrease in irradiance, rubisco deactivated slowly for at least 25 minutes. Bound RuBP increased gradually but did so more slowly than the estimated increase in inactive rubisco sites.