Two experiments using a lexical priming paradigm investigated how stress information is processed in reading Italian words. In both experiments, prime and target words either shared the stress pattern or they had different stress patterns. We expected that lexical activation of the prime would favour the assignment of congruent stress to the target. Results showed that participants were faster in naming target words that had the same stress pattern as the prime. Similar effects were found on target words that were included in lists in which all prime and target stimuli had the same stress type (Experiment 1) and in lists with mixed stress type and congruency between primes and targets (Experiment 2). Results indicate that, in single word reading, metrical information about stress position is activated in the lexicon, independent of segmental information. (Contains 4 tables and 3 footnotes.)