Poor automatization of lexical access results in poor performance in reading comprehension (Perfetti, 1985 and 1991). This study relates the performance of 84 undergraduate students in word recognition and reading comprehension skills. Three items of a standardized test were given to the students—the reading of words and pseudowords to identify the percentage of error, and a reading comprehension item. The results show that 36% had high percentages of error in both lexical and semantic processes, while 19% had low percentages of error in both tasks. The effect of the lexical variables of frequency and length is evident. There is a higher percentage of error for low-frequency words and pseudowords and for long words and pseudowords. A higher rate of error in words and pseudowords correlated to a greater number of mistakes in reading comprehension.