Consider a set of players that are interested in collectively evaluating a set of objects. We develop a collaborative scori ng protocol in which each player evaluates a subset of the objects, after which we can accurately predict each players’ individual opinion of the remainingobjects. The accuracyof thepredictionsisnearoptimal,depending onthenumberof objects evaluated by each player and the correlation among the players’ preferences. A key novelty is the ability to tolerate malicious playe rs. Surprisingly, the malicious players cause no (asympt otic) loss of accuracy in the predictions. In fact, our algor ithmimprovesinbothperformance and accuracy overprior state-of-the-art collaborative scoringprotocolsthatprovided no robustness to malicious disruption.