Three experiments examined the effects of unconsciously presented picture primes on semantic categorization and naming responses to both word and picture targets. Picture naming and word categorization responses to targets were faster and more accurate when the picture primes belonged to the same semantic category as the targets, so called priming effect. No priming was found when subjects performed a word reading task. When priming was evident, no difference was found between responses to targets that were nominally identical to primes (e.g. the picture of a lion followed by either the word LION or the picture of a lion) compared with nominally different targets from the same semantic category as the primes (e.g. the picture of an ELEPHANT followed by either the word LION or the picture of a lion). Responding did not differ significantly from chance when subjects were asked to categorize the primes as natural objects vs. artifacts or as meaningful vs. meaningless objects in three distinct forced-choice unspeeded tasks.