We suggest the measurement of Drell-Yan (DY) lepton pairs in p-Pb collisions at the LHC ( √ s = 5.02 TeV) in order to disentangle the relative contributions of shadowing and coherent energy loss in quarkonium production off nuclei. The nuclear modification of low mass DY production is computed at NLO using various sets of nuclear parton densities. It is then observed that shadowing effects strongly cancel out in the J/ψ over DY suppression ratio R ψ pA (y) R DY pA (y), unlike the effect of coherent energy loss. Such a measurement could be performed at forward rapidity by the ALICE and LHCb collaborations at the LHC. PACS numbers: 24.85.+p, 14.40.Pq, Measurements of J/ψ production in p-Pb collisions at the LHC ( √ s = 5.02 TeV) by ALICE [1,2] and LHCb [3], and the observation of a strong attenuation at large rapidity with respect to the p-p data interpolation at the same collision energy, has triggered an intense debate on the origin of such a nuclear suppression [4]. Several groups have attributed the suppression to the depletion of the gluon distribution in the target nucleus expected at small x 2 10 −2 . Such a depletion, commonly named 'shadowing', is currently either incorporated in nuclear parton distribution functions (nPDFs) obtained from global fits based on DGLAP evolution, or determined from non-linear QCD evolution within the saturation formalism (see [5,6] for topical reviews). However another fundamental phenomenon, namely coherent parton energy loss in cold nuclear matter [7][8][9][10][11][12][13][14], could explain these data. It is therefore crucial to quantify the relative effects of shadowing and coherent energy loss. In this Letter, we show that the measurement of Drell-Yan (DY) lepton pairs of relatively low mass (10 M DY 20 GeV) in p-A collisions could be decisive to reach this goal. As a consequence, this measurement would shed light on the production of hadrons in p-A collisions, expected to be sensitive to coherent energy loss [7] on top of possible shadowing effects.Coherent energy loss and shadowing are two distinct effects, and should in principle be both taken into account in nuclear suppression models. However, as discussed in [8, 10] coherent energy loss alone allows one to describe J/ψ nuclear suppression observed at large x F at fixed-target collision energies, √ s 40 GeV [15], with a value of the cold nuclear matter transport coefficient q 0 = 0.07 − 0.09 GeV 2 /fm. This is consistent with the fact that shadowing is expected to be small at those energies. Taking in addition into account the shadowing given by either EPS09 [16] or DSSZ [17] next-to-leading order (NLO) nPDF sets these data can be described with a slightly smaller value ofq 0 , although energy loss re-mains the dominant effect at fixed-target energies [8]. Extrapolating coherent energy loss to collider energies leads to a central prediction [8] (with a rather narrow theoretical uncertainty band [4]) which agrees well with the J/ψ suppression data measured in d-Au collisions at RHIC [18,19] and p-Pb collisions at the ...