The standard pseudohomophone effect in the lexical decision task, i.e. longer response times and higher error rates for pseudohomophones compared with spelling controls, is commonly explained by an orthography-phonology-conflict. This study tested this conflict account, using a multi-method approach including participant's behavioral responses, confidence ratings, pupillary responses and event-related potentials (ERPs). The classic pseudohomophone effect was replicated using relatively long, multi-syllabic stimuli. Pseudohomophones were rated less confidently as being nonwords than spelling controls, and they affected the pupillary response by increasing the peak pupil diameter. Both findings are interpreted in terms of increased conflict and higher cognitive demands leading to uncertainty while solving the task. The ERP revealed an N400 component for spelling controls, showing a graded effect: word<pseudohomophone<spelling control. This can be seen as evidence for (partial) semantic activation through pseudohomophones. Taken together, the results provide strong multi-method evidence for the conflict account of the pseudohomophone effect.