Three experiments examined whether processes devoted to word recognition and word identification are suppressed during saccades, as most eye movement and reading researchers implicitly assume. In the first two experiments, subjects made short or long saccades while performing lexical decisions; lexical decision latency and accuracy were unaffected by saccade distance, and post-saccadic processing time was reduced when a long as opposed to a short saccade was made. Experiment 3 showed that word identification is more accurate when a long as opposed to a short saccade separates the presentation of a word and the presentation of a mask. These results demonstrate that lexical processing is not suppressed during saccades, so saccade durations should be taken into account in eye movement studies of reading. The implications of the results for current theories of cognitive suppression during saccades are discussed.