In order to have a better understanding of the nature of reading behavior, it is quite essential to scrutinize the underlying assumptions of the two competing theories: the componential approach and unitary hypothesis and see how this is reflected on testing the reading skills. As a result, the current study was aimed at broadening the existing knowledge horizon by trying to identify if there was relationship in the overall scores of learners at different ability groups in Reading Sub-skills Test and reading levels in Cloze Test and showing how they differed in their performance with respect to specific reading sub-skills. So as to address the specific research objectives, tests were used to collect the required data. More specifically, these included: a cloze test to determine the subjects’ reading levels; and a reading sub-skills test to find out their performance in terms of the sub-skills at word, sentence, and discourse level. Descriptive statistics (i.e. frequencies and percentages) and correlations were used to analyze the data collected. Besides, the SPSS was used to carry out the computations and normality tests which showed that the test scores were not normally distributed. Hence, Kendall’s tau, a non-parametric test, was used to carry out correlation coefficients. In addition, the data analysis revealed that average-scorers, low-scorers, test-takers at instructional reading level, and those at frustration level had much difficulty with most of the reading sub-skills with varying degrees. Eventually, recommendations were suggested so that learners could improve their difficulties with the reading sub-skills.