Intermittent heating is to maintain indoor temperature of buildings in winter within a comfort range only during occupancy periods. Buildings have thermal inertia causing slow increments of indoor temperature, so that a preheating time is required to start heating in advance of occupancy periods. In this paper, a new method is presented to determine preheating time for building intermittent heating subject to changes in outdoor temperature and solar radiation. First, a dynamic model is developed to describe relations between indoor temperature as an output and three inputs including the heating medium temperature, outdoor temperature, and solar radiation. The dynamic model with modeling uncertainties is identified from input-output historical data. Second, the identified model is used to predict indoor temperature under future data of three inputs, and optimal preheating times are determined with a certain probability to ensure successful preheating. The experimental examples show that the dynamic model accurately describes the indoor temperature changes, and the preheating times from the proposed method are more accurate than four existing methods.