Selection operates at many levels. Robert Brandon has distinguished the question of the level of selection from the unit of selection, arguing that the phenotype is commonly the target of selection, whatever the unit of selection might be. He uses "screening off" as a criterion for distinguishing the level of selection. Cave animals show a common morphological pattern which includes hypertrophy of some structures and reduction or loss of others. In a study of a cave dwelling crustacean, Gammarus minus, we (...) find evidence for selection for both increased antennal size and reduction of eyes. The genetic structure of the population does not support the view that the phenotype screens off the genotype in explaining the differences in fitness. Nonetheless, the results do indicate that the level of selection is at least at the level of the phenotype in both cases. (shrink)