The universal baryonic mass fraction (Omega(b)/Omega(m)) can be sensitively constrained using X-ray observations of galaxy clusters. In this paper, we compare the baryonic mass fraction inferred from measurements of the cosmic microwave background with the gas mass fractions (f(gas)) of a large sample of clusters taken from the recent literature. In systems cooler than 4 keV, fgas declines as the system temperature decreases. However, in higher temperature systems, f(gas)(r(500)) converges to approximate to(0.12 +/- 0.02) (h/0.72)(-1.5), where the uncertainty reflects the systematic variations between clusters at r(500). This is significantly lower than the maximum-likelihood value of the baryon fraction from the recently releasedWilkinson Microwave Anisotropy Probe (WMAP) 3-yr results. We investigate possible reasons for this discrepancy, including the effects of radiative cooling and non-gravitational heating, and conclude that the most likely solution is that Omega(m) is higher than the best-fitting WMAP value (we find Omega(m) = 0.36 +(+0.11)(-0.08) , but consistent at the 2 sigma level. Degeneracies within the WMAP data require that sigma(8) must also be greater than the maximum likelihood value for consistency between the data sets.