This study aims at providing a comprehensive account of the types of errors produced by Arab students of English as a second language based on a multiple classificatory taxonomy developed for this purpose. The corpus providing the database for the study consists of three parts: (i) short tape-recorded interviews, (ii) translated sentences and (iii) writing a one-page essay. The subjects of the study, who were by then attending the Intensive English Language Institutes (IELI) at the University of Texas at Arlington and University of North Texas, belong to four Arab countries and, thus, they represent different dialect, educational and socioeconomic backgrounds. Irrespective of the variations attributed to the dialect differences, which are beyond the scope of the present paper, findings have shown that: (i) intralingual errors are the most commonly occurring types, (ii) lexico-semantic errors, which are mainly interlingual, contribute substantially to the grand percentage of errors, representing (39.22%) of all linguistic errors, and (iii) the majority of errors affect only one item in the utterance, and thus, they never lead to miscommunication between the sender and the receiver. (Contains 2 tables.)