The decision of using English as medium of instruction for mathematics and science subjects in public sector schools of the Punjab province of Pakistan invokes serious issues regarding its potential impacts on teacher effectiveness and student achievement. So far as the teachers are concerned, perhaps one of the best documented attributes of effective teachers is a strong sense of efficacy (Henson, Kogan, & Vacha-Haase, 2001). The study at hand was designed to explore the self-efficacy beliefs of school teachers make a comparison on the basis of medium of instruction. For comparison purpose, the sample consisted of teachers who taught mathematics and/or science any of the classes that had been declared to be English medium by the government. These classes generally include class 1,2, and 6. In some schools where this innovation had been introduced earlier, however, class 3 and 7 were also English medium. Present study, intended for comparison, is descriptive in nature and the method used was a survey research. Multistage stratified random sampling technique was used to draw sample from the population of school teachers from all public-sector primary, middle, and secondary schools of the Punjab province. Sample size was 1761 with 880 male teachers and 881 female teachers, 861 Urban and 900 rural teachers, 923 English medium and 838 Urdu medium teachers selected from 419 randomly selected primary, middle and high schools of the nine districts selected from nine administrative divisions of the Punjab province randomly. Data were collected through Teachers’ Sense of Efficacy Scale (TSES) developed by Tschannen-Moran & Woolfolk (2001). Long form of this viii questionnaire having 24 items on three different sub scales was used. The statements were translated into Urdu through translation-back-translation method with the help and guidance of a panel of experts. The bilingual version of the questionnaire was used to collect data. The collected data were analyzed using independent samples t-test, two way analysis of variance, and multivariate analysis of variance. The study revealed that Urdu medium school teachers had a higher level of self-efficacy as compared to English medium teachers on overall TSES scores as well as on the three sub-scales of the instrument. Gender was not found significant for efficacy difference. School level, school local, designation and teaching experience wise difference reached the level of significance. The findings of the study invite for a deeper probing into the matter and rethinking over the recently taken decision of transforming school education system form Urdu to English medium.