To investigate phonological effects in visual word recognition, a visual lexical decision task was used in which the critical stimuli were two types of highly familiar Hebrew acronyms: 'phonological' acronyms that are conventionally pronounced as single words via letter-to-sound translation (?"? (d"?) = /da?/), and 'lexical' acronyms that are conventionally pronounced according to their full multiword name (?"?(t"?) = /tel/ /aviv/). Thus, in the case of 'phonological' acronyms, phonological recoding may contribute to the recognition process, while in the case of 'lexical' acronyms, it may interfere ((?"?t"?) ? ta?/). If familiar letter strings are accessed mainly orthographically, as assumed by dual route models, then no difference is expected between these two types of acronyms. Alternatively, if phonological recoding influences word recognition, then 'phonological' acronyms should be easier to recognise. Consistent with this latter interactive-connectionist view, responses were faster and more accurate in the 'phonological' than in the 'lexical' condition.