A large sample of X-ray-emitting clusters of galaxies is considered in the context of Milgrom's modified Newtonian dynamics (MOND). The virial discrepancy (the ratio of the dynamical to observable mass) is reduced, on average, from a factor of 4 in the Newtonian analysis to a factor of 2 in the MOND analysis. The reason for this modest reduction is that the internal accelerations in the inner 1 Mpc of the sample clusters are not very small in terms of the critical MOND acceleration: a approximate to 0.5a(0). Although the uncertainty is large, there remains a discrepancy that is not yet accounted for by the directly observed matter in galaxies and diffuse gas. This may be consistent with the fact that modified dynamics cannot resolve the strong lensing discrepancy in those clusters where this phenomenon occurs. If MOND is correct, then the implication is that the budget of possibly detectable matter is not yet complete, particularly in the central regions of rich clusters.