We infer the crater chronologies of Ceres and Vesta from a self-consistent dynamical model of asteroid impactors .
The model accounts for planetary migration/instability early in the solar system history and tracks asteroid orbits over 4.56 Gy .
It is calibrated on the current population of the asteroid belt .
The model provides the number of asteroid impacts on different worlds at any time throughout the solar system history .
We combine the results with an impactor-crater scaling relationship to determine the crater distribution of Ceres and Vesta and compare these theoretical predictions with observations .
We find that : ( i ) The Ceres and Vesta chronologies are similar , whereas they significantly differ from the lunar chronology .
Therefore , using the lunar chronology for main belt asteroids , as often done in previous publications , is incorrect .
( ii ) The model results match the number and size distribution of large ( diameter > 90 km ) craters observed on Vesta , but overestimate the number of large craters on Ceres .
This implies that large crater erasure is required for Ceres .
( iii ) In a model where planetary migration/instability happens early , the probability to form the Rheasilvia basin on Vesta during the last 1 Gy is 10 % , a factor of \sim 1.5 higher than for the late instability case and \sim 2.5 times higher than found in previous studies .
Thus , while the formation of the Rheasilvia at \sim 1 Gy ago ( Ga ) would be somewhat unusual , it can not be ruled out at more than \simeq 1.5 \sigma .
In broader context , our work provides a self-consistent framework for modeling asteroid crater records .