To determine the number and types of spelling errors made by college students, a study examined students' spelling errors on four complex word patterns which occur in related words. Fifty students attending a four-year engineering institute were selected and classified as group A (good readers/good spellers) or group B (poor readers/poor spellers) based on recommendations from their English professors and by a Spache-based spelling test. Reading ability was determined by Nelson Denny Test, Form C, 1973. Students were asked to spell words occurring in pairs with a change in the quality of the pronunciation of a vowel or consonant and/or stress pattern of the derivative, (e.g., vowel reduction in "compete"/"competition," or alternation from silent to articulated consonant in"malign"/"malignant"). Scoring focused on the critical grapheme or graphemic sequences preserving the relationship between the pairs, and classified errors as either phonetic or nonphonetic. Results show that group A had mastered the spelling of difficult words, and had internalized phonological rules governing predictable changes in word pairs; group B had not, presumably because of their limited familiarity with print. The few errors made by group A were generally phonetic, while group B's were overwhelmingly nonphonetic. Group A appeared to spell on a lexical rather than a phonological level; group B relied on phonological information which is usually ineffective when spelling polysyllabic words with silent letters and changes in stress patterns. (JG)