This paper reviews 150 years of progress towards understanding the succession of Pleistocene ice ages. Emphasis is placed on the process of explaining forced variations in climate which occur in the Milankovitch band (periods from 10000 to 400 000 years), where astronomical forcing functions are clearly identified and where the amplitudes of climatic change are large. An hierarchy of explanatory models is employed, including a statistical model that uses gain and phase terms to parameterize the climatic response to orbital forcing in several narrow frequency bands. This type of model accounts for a substantial fraction of the observed temporal variations of δ18O over the past 800 000 years. Experiments using the model suggest that a change in the system response occurred about 400 000 years ago. The problem of explaining climatic variations in other frequency bands is briefly reviewed. In the decadal band (l0–400 years), there is good evidence of volcanic forcing, solar forcing, and tidal forcing, but much of the observed pattern is presumed to reflect free variations. In the millenium band (400 to 10 000 years), significant climatic changes occur at periods ranging from 1000 to 3000 years. Their cause remains a challenging problem. Over the tectonic band (more than 400 000 years), volcanic fluxes, continental elevation, and continental position are the most likely forcing functions of the observed climatic changes.