Core body temperature (CBT) is a basic physiological parameter that reflects the level of human metabolism and is therefore clinically important for the diagnosis of various diseases. The traditional method of CBT evaluation based on heat flux is susceptible to errors caused by horizontal heat flux and the sensor is usually large in size. In this study, a novel miniaturized sandwich-like CBT monitoring sensor (SCBTS) based on negative temperature coefficient (NTC) temperature sensor and heat flux sensor was constructed. The sandwich-like structure minimized the effect of horizontal heat flux and reduced ambient temperature disturbance. In addition, a novel numerical method was explored to predict CBT during unsteady heat flux. Effectiveness of SCBTS was verified by finite element method (FEM) simulation, mock-up experiments and human experiments. The proposed sensor was effective within the range of human body temperature fluctuations (35°C–42°C) with an average error of 0.02°C on the hot plate. In human trials, resting experiments (eight participants for 13 trials) showed 0.02°C average difference with a 95% CI of [−0.16, 0.12]°C. Furthermore, the sensor showed distinct temperature changes in different periods using exercise experiments (6 participants). The proposed SCBTS continuously monitors temperature without causing inconvenience to the subject, therefore, it is highly suitable for application.