The aim of the study was to evaluate the construct validity and psychometric properties of two short scales intended to measure Current Academic Learning Self-Efficacy (CAL-SE) and Current Academic Exam Self-Efficacy (CAE-SE) using Rasch measurement models, and to investigate whether these differed across four degree programs. Data consisted of 643 Danish undergraduate students two universities. Item analyses showed both the CAL-SE and the CAE-SE scales to fit graphical loglinear Rasch models. In the CAL-SE scale two items were locally dependent. In the CAE-SE scale, two item pairs were locally dependent, and one item functioned differentially relative to degree program. Reliability of both scales was satisfactory for statistical purposes with relatively large samples. Targeting was good for both scales. Economy students scored significantly lower than the joint group of law, psychology and political science students, both with regard to current academic learning self-efficacy and current academic exam self-efficacy.