Abstract:  The meritocratic fairness ideal implies that inequalities in earnings are regarded as fair when they reflect differences in performance, but not when they reflect differences in luck. A straightforward implementation of the meritocratic fairness ideal requires complete information about individual performance, however, and such information is often unavailable. We study the highly realistic, but previously under-studied, context where there is uncer­tainty about whether a particular inequality reflects performance or luck. We show theoreti­cally that meritocrats in such situations may either be pulled towards an egalitarian behavior, or stay far away from it. The direction in which an individual meritocrat is pulled de­pends on how she trades off the two main errors that can arise when redistributing under uncertainty – the first error being that the worst performer instead of the best is rewarded, and the second being that whereas the best performer is rewarded, she is not rewarded enough. A laboratory experiment shows results that are well aligned with our model, revealing that it is significantly more common for meritocrats to be pulled towards (as opposed to staying away from) an egalitarian style behavior under uncertainty. The external validity of our framework, and the results from the laboratory, are confirmed in two large-scale surveys conducted on representative samples of the American and the Norwegian populations