In metal cutting, the ability to model its tool temperature distribution is highly desirable, as it provides an effective means to monitor the tool and workpiece conditions, particularly when machining hard-to-machine materials that have a low heat conductivity. Because of the complex heat generation in the microscale tool–chip interface, the difficulty to infer its temperature distribution is a well-known problem. In the context of dry lathe turning, this paper presents a hybrid method that considers both the macroscale tool heat transfer and microscale machining mechanics to reconstruct the 3-D tool temperature field from nonobstructed infrared (IR) images. The microscale-mechanics model identifies the contact geometry and estimates the frictional heat input to determine the complete boundary conditions to solve the macroscale heat-transfer model for the steady-state 3-D temperature distribution that provides a basis for experimentally fitting the model by comparing the computed surface temperature with actual temperature measurements. Two sets of experimental results validating the reconstructed temperature on a customized orthogonal-cutting testbed with a high-resolution IR imager and evaluating the effectiveness of the hybrid method on a lathe-turning center are presented. The results demonstrate that with the hybrid macro–micro modeling, the 3-D steady-state temperature field of a typical commercial lathe tool insert can be accurately reconstructed from a relatively low-resolution IR image.