Milankovitch theory-that climate is controlled by variations in the Earth's orbital parameters-has gained wide acceptance for its ability to account for two climate cycles: a 23-kyr cycle that is phase-locked to the precession-driven insolation cycle, and a 41-kyr cycle that is phase-locked to the obliquity-driven insolation cycle. But, explaining the observed ~100- kyr climate cycle in terms of Milankovitch theory-especially for the Late Pleistocene ice-age cycle-remains controversial in spite of a strong correlation with the ~100-kyr cycle in the Earth's orbital eccentricity. One problem is that eccentricity affects insolation mainly by modulating the precession cycle; its direct contribution to radiation change is too small (<0.1%) to effect the observed climate change directly. Another is the absence of a Late Pleistocene ice-volume cycle in oxygen-isotope records to match the ~404-kyr component of the eccentricity cycle. Here we examine an oxygen-isotope record spanning the interval 1.2 to 5.2 million years ago, before the Late Pleistocene ice-age regime. We find 404-kyr and ~100-kyr climate cycles which are coherent with eccentricity and which have amplitudes that are similar to the coexisting 23-kyr cycle. Analysis of these low- frequency cycles suggests that they originate through an asymmetrical response mechanism that preferentially introduces variance into the climate system from the warmer portions of the eccentricity-modulated precession cycle. Our data thus support eccentricity's role in the origin of low- frequency oxygen-isotope cycles before the Late Pleistocene ice age.