Numbers (i.e. 45) are symbols that are common in reading texts in various forms such as dates, percentages, and quantity expressions. Each digit corresponds to a word, unlike words in which each letter represents a phoneme. Despite their high frequency in reading texts, eye movement research on numeral processing is still rare. Numerals are easy to process in silent reading due to their standard nature in many languages; however, in oral reading, extra effort is required as numerals are pronounced irregularly among several languages. This study aimed to examine processing differences between words and numerals during second language oral reading by using eye-tracking. 40 intermediate non-native speakers were tested under same three text conditions (familiar words, non-words, and numerals) while their eye movements were recorded. Results showed that the participants spent a considerable amount of time processing numerals when compared to words and non-words. Additionally, numerals caused more regressions and were revisited more causing slower reading performance.