Conflicting evidence exists about the extent of melt-mantle reaction as melt ascends beneath a mid-ocean ridge, On the one hand, uranium series (U-series) disequilibria in mid-ocean ridge basalts suggest that melts reactively flow towards the surface and continuously re-equilibrate with lower pressure mantle. On the other hand, trace elements in abyssal peridotites suggest fractional melting and isolation of ascending melt from surrounding lower pressure peridotite. Following a suggestion by Kelemen ct al. (1997) [Kelemen, P.B., Hirth, G,, Shimizu, N., Spiegelman, M., Dick, H.J.B,, Philos. Trans. R. Soc. 355, 283-318], I construct a model for melting beneath mid-ocean ridges which consists of two materials having different melt porosity distributions in order to quantitatively assess whether the U-series and abyssal peridotite observations could actually be compatible. The melting column consists of a volumetrically dominant lherzolite which produces melt by decompression and a volumetrically minor dunite which serves as a conduit for fast melt ascent. A suction parameter S (after Iwamori (1994) [Earth Planet. Sci. Lett, 125, 1-16]) controls the transfer of melt between the Iherzolite and dunite media. S can vary from 0 where none of the ascending melt flows through the dunite melt conduit (reactive porous flow melting) to 1 where the melt produced in the iherzolite is immediately transferred to the dunite (near-fractional melting). Transport times of U-series nuclides are accounted for in both porosity regimes based on calculated flux balances. Results show that the trace element patterns observed in abyssal peridotites require S to range from 0.5 to 0.8 depending on assumptions about the composition of the original source material. U-series models are evaluated in terms of three quantities: S, chi (the volumetric fraction of dunite), and a(2)/etaC which lumps together several constants relating to permeability. Ra-226 and Pa-231 excesses are most sensitive to a(2)/etaC (which affects the melt porosity within the melting region). Ra-226 excesses match the observed levels of Ra-226 excess in mid-ocean ridge basalts (MORB) at all values of S. Pa-231 excesses favor lower values of S as models with high S produce significantly less (Pa-231)/(U-235) than that observed. There remains some disparity between the values of S required by the abyssal peridotite data and the observed Pa-231 excesses although further refinement of partition coefficients could resolve the difference. Within the models presented, the two geochemical observations are close to being reconciled if S=0.5 and the mantle permeability is at least 1 x 10(-14) m(2). However, a factor of two higher bulk partition coefficient for U (D-U=0.003) would allow the observed ((Pa-231)/(U-235)) to be matched at S=0.5 while dropping the required bulk permeability to less than 1x10(-15) m(2). (C) 2000 Elsevier Science B.V. All rights reserved.