Response priming occurs when prime and target are members of the same category in categorisation tasks. Prior research has demonstrated these effects, e.g. in number magnitude categorisation tasks, but has failed to establish a congruency priming effect in masked primed lexical decision. We present a series of five experiments that demonstrate clear prime congruency effects in lexical decision. The data show that the size ofthe congruency priming effect is influenced by the difficulty of the word-nonword discrimination. This sensitivity can explain the failure to find prime congruency effects in previous studies. Furthermore, the data suggest that the lexical decision is dependent on an evidence accumulating process, allowing the prime to bias speed and accuracy. We discuss the implications of these results for computational models of visual word identification and lexical decision.