Ras GTPase activating proteins (GAPs) contain an invariant motif, -FLR-, within the most conserved region of their catalytic domains. Certain mutations in this motif have greatly reduced activity (Skinner, R. H., Bradley, S., Brown, A. L., Johnson, N. J., Rhodes, S., Stammers, D. K., and Lowe, P. N. (1991) J. Biol. Chem. 266, 14163-14166), but it was not determined whether the reduced activity was due to loss of binding or impaired catalysis. In order to address this question, we have developed a simple physical method to study formation of GAP.p21ras complexes. This utilizes the increase of fluorescence anisotropy upon binding of GAP to p21ras complexed with 2'(3')-O-(N-methylanthraniloyl) (mant) derivatives of guanine nucleotides. Dissociation constants obtained for the catalytic domains of either p120-GAP (GAP-344) or neurofibromin (NF1-GRD) with normal and Leu-61 p21ras proteins are comparable with those obtained by kinetic methods. In the course of these studies, we found, in contrast to previous observations, that both GAP and NF1-GRD can weakly activate the GTPase of Leu-61 mutant p21, showing that Gln-61 is not absolutely required for the stimulation of GTPase activity by GAPs. The fluorescence anisotropy method allowed us to show that mutation of Arg-903, within the FLR motif of GAP, can result in protein defective in catalysis but not in binding to p21ras. These data suggest a direct role for this residue in catalyzing GTP hydrolysis on p21ras, possibly by contributing a catalytic group to the p21 active site. This method is independent of the catalytic activity of the proteins, and so it could be extended generally to the measurement of binding of effector molecules, exchange factors, or other macromolecules to guanine nucleotide-binding proteins.