Pyruvate kinase (PK, EC 2.7.1.40) was partially purified from the plant cytosolic fraction of N2-fixing soybean (Glycine max [L.] Merr.) root nodules. The partially purified PK preparation was completely free of contamination by phosphoenolpyruvate carboxylase (PEPC, EC 4.1.1.31), the other major phosphoenolpyruvate (PEP)-utilizing enzyme in legume root nodules. Latency experiments with sonicated nodule extracts showed that Bradyrhizobium japonicum bacteroids do not express either PK or PEPC activity in symbiosis. In contrast, free-living B. japonicum bacteria expressed PK activity, but not PEPC activity. Antibodies specific for the cytosolic isoform of PK from castor bean endosperm cross-reacted with a 52-kDa polypeptide in the partially purified PK preparation. At the optimal assay pH (pH 8.0 for PEPC and pH 6.9 for PK) and in the absence of malate, PEPC activity in crude nodule extracts was 2.6 times the corresponding PK activity. This would tend to favour PEP metabolism by PEPC over PEP metabolism by PK. However, at pH 7.0 in the presence of 5 mM malate, PEPC activity was strongly inhibited, but PK activity was unaffected. Thus, we propose that PK and PEPC activity in legume root nodules may be coordinately regulated by fluctuations in malate concentration in the plant cytosolic fraction of the bacteroid-containing cells. Reduced uptake of malate by the bacteroids, as a result of reduced rates of N2 fixation, may favour PEP metabolism by PK over PEP metabolism by PEPC.