The present study concentrates on the organization of the mental lexicon with regard to semantic transparency in the representation of Polish compounds. Its aim was to test current approaches to the processing of morphologically complex words in a lexical decision experiment with the use of visually presented Polish compound and simple words. The existing psycholinguistic approaches centre around the same question: are complex words parsed into their constituent parts or are they stored as full-word representations in the human mental lexicon? I referred to five widely acknowledged models of morphological processing to account for the outcomes of the present study. The data reveal that: (i) transparent compounds primed by words semantically related to the heads of these transparent compounds elicited faster response times than opaque compounds within the same condition; and (ii) priming speeds up the processing for both transparent and opaque compounds. The results indicate that the processing of Polish compound words is influenced by semantic transparency and that both transparent and opaque compounds are decomposed into their constituents prior to lexical access.