Abstract. We use a statistical model, the cointegrated vector autoregressive model, to evaluate the relative roles that orbital, seasonal, and spatial variations in solar insolation play in glacial cycles during the late Quaternary (390kyr – present). To do so, we estimate models of varying complexity and compare the accuracy of their in-sample simulations. Results indicate that variations in solar insolation associated with changes in Earth's orbit have the greatest explanatory power and that obliquity, precession, and eccentricity are needed to generate an accurate simulation of glacial cycles. Seasonal variations in insolation play a lesser role, while cumulative summer-time insolation has little explanatory power. Finally, solar insolation in the Northern Hemisphere generates the more accurate in-sample simulation of surface temperature while ice volume is simulated most accurately by solar insolation in the Southern Hemisphere.