This paper introduces the type II signal detection model as a means of modeling metacognitive sensitivity of foreign language learners. Foreign language teaching researchers have paid much attention to the conceptual aspects of metacognitive sensitivity, but not to its mathematical formulation and measurement models, while cognitive and mathematical psychologies have voluminous literature on how to model judgment behaviors mathematically. Among mathematical models of judgment behaviors, this paper particularly focuses on the type II signal detection model, one of the most dominant practices of modeling metacognitive sensitivity. The model is nothing but an application of the equal-variance Gaussian signal detection model to type II tasks such as trial-by-trial binary confidence rating during a stimulus discrimination task, which usually corresponds to type I tasks. The model basically calculates two indices, (a) type II d’ or metacognitive sensitivity and (b) type II biases or metacognitive biases, based on type II response categories below: (a) type II hit, (b) type II false alarm, (c) type II miss, and (d) type II correct rejection. By reviewing such mathematical basics and showing a computation of the model in a Bayesian manner with some numerical examples, this paper claims that the type II signal detection model can be a potentially beneficial tool for foreign language teaching researchers.