This study examined the neighborhood characteristics (frequency, density, and neighborhood frequency) of 138 malapropisms (whole word errors). A qualitative analysis suggests that words that are more often involved in malapropisms have low neighborhood frequencies. Additionally, these words tend to be either high-frequency target words in dense neighborhoods, or low-frequency target words in sparse neighborhoods. Finally, malapropisms tended to occur such that the word that was substituted for the intended word was more frequent than the intended word. The implications of these findings for models of lexical representation and processing are discussed.