The question of how word morphology is coded and retrieved during visual word recognition has given rise to a large number of empirical studies. The results, however, do not enable one to decide between alternative models of morphological representation and processing. It is argued in this paper that the contrast between pseudoprefixed words and non-prefixed control words can provide an empirical basis for deciding between hypotheses of morphology representation as sublexical or lexical. This contrast has been used in the three lexical decision experiments reported here, which show that decision times for pseudoprefixed words are significantly slower than for non-prefixed control words. This pseudoprefixation effect strongly supports the hypothesis that morphology is coded and processed sublexically during word recognition. The experimental conditions employed allow both strategic and strictly orthographic explanations for the pseudoprefixation effect to be dismissed.