U This study investigated spelling errors in compositions written by students from four different language backgrounds at Iowa State University. Fifty-six writing samples were collected: nine from Arabic speakers, ten from Chinese speakers, twenty from Malay speakers, and seventeen from Spanish speakers. All of the students had scored 80-89 (inclusive) on the Michigan Test of English Language Proficiency and all had achieved a minimum of 500 on the TOEFL. They were all enrolled in academic programs at Iowa State during the 1981-1982 school year. Four hypotheses were tested: 1) that ESL students at this proficiency level would make more spelling errors than native speakers, 2) that the ESL students would make more habitual errors (real mistakes) than slips (misspellings which are corrected elsewhere in the same composition), 3) that at this proficiency level there would be more spelling errors among the speakers of languages that use the Roman alphabet (Spanish, Malay) than among the speakers of languages that do not (Arabic, Chinese), and 4) that ESL students would err more frequently in the medial position of the word than initially or finally. The results showed that for the four language groups the error percentage mean was 1.88% of total words written; native speakers exhibit a 1.1% error rate (Chedru and Gerschwind, as cited in Wing and Baddeley 1980). Second, the research showed that the "slip" mean for all languages was .19 errors, while the habitual error mean was 3.66 errors. Third, contrary to Oller and Ziahosseiny's findings (1970), the difference between the nonRoman and the Roman alphabet language error percentages was found to be insignificant. Fourth, it was found that the means for error position of all four groups were as follows: initial was .428 errors, medial was 2.61 errors, and final was .66 errors.