The study investigated the effect of problem-based learning through cognition-based tasks on speaking proficiency of Iranian intermediate EFL learners in comparison to the effect of objective-based tasks. To this end, a true experimental research design was employed. Ninety five (N = 95) language learners studying at a language institute in the city of Esfahan, Iran were given an IELTS listening and speaking test as the proficiency test and 75 learners were selected. In the next phase of the study, a second IELTS speaking test was administered as the homogeneity test and the pre-test to seventy five (N = 75) learners chosen from the population and forty-eight (N = 48) homogeneous intermediate learners were selected for the study (i.e., 24 learners in control group and 24 in experimental one). The results of an independent-sample t-test gained from the study proved that not only does implementation of problem-based learning through cognition-based tasks significantly increased intermediate participants' speaking proficiency, but also it had more positive effect in comparison to objective-based tasks on participants' speaking proficiency. Therefore, it is suggested that problem-based learning ought to be taken into account by educational scholars, those in charge of syllabus, material producers, language teachers and language learners.