Back-arc melting regimes provide the unique opportunity to enhance our knowledge of the influence of mantle heterogeneity on mantle melting, i.e. of the extent of arc related, fluid-induced melting, vs. decompressional melting. However, few U-Th-Ra disequilibria have been analysed in back-arc lavas so far. Manus back-arc basin lavas range in composition from NMORB/ transitional MORB to back-arc basin basalts (BABB) and some extremely enriched BABBs (XBABB). We have analysed U-Th-Ra disequilibria 100-400 km off the active New Britain trench. While most display (230Th/238U) ratios of 0.77-0.95 a few lavas both closer and more distant to the trench have small Th-excesses with (230Th/238U) <1.06. (226Ra/230Th) ratios are generally >1 implying that the lavas are <8 ka old and that no post-eruptive age correction needs to be applied to U-Th. While (230Th/238U) >1 are generally associated with decompressional melting, (230Th/238U) <0.95 have generally been attributed to the presence of arc-related fluids or unusual melting phases (e.g. apatite , however, we do not find evidence for apatite in these rocks). The Sr-Nd-Pb isotopes imply variable input of an enriched component, i.e. they range from ‘MORB’-like depleted signature to more enriched signatures. There is no evidence for a correlation of arc-related melting signatures and long-term source heterogeneity. Instead the U-Th disequlibria are consistent with fluid-induced melting some 200-400 km distant from the active trench. This U-excess requires a) unusual phases during melting or b) storage of fluids in the mantle and re-melting (e.g. amphibole, however, this would lead to Th-rather than U-excess). We thus favour a much quicker/further transport of the arc-related fluid through the mantle wedge (with or without intermediate storage) inducing melting much further away from the active trench then suggested by dehydration of the subducted slab.