The research of exoplanets has entered an era in which we characterize extrasolar planets. This has become possible with measurements of radii and luminosities. Meanwhile, radial velocity surveys discover also very low-mass planets. Uniting all this observational data into one coherent picture to better understand planet formation is an important, but difficult undertaking. Our approach is to develop a model which can make testable predictions for all these observational methods. We continue to describe how we have extended our formation model into a self-consistently coupled formation and evolution model. We show how we calculate the internal structure of the solid core and radiogenic heating. We also improve the protoplanetary disk model. Finally, we conduct population synthesis calculations. We present how the planetary mass-radius relationship of planets with primordial H/He envelopes forms and evolves in time. The basic shape of the M-R relation can be understood from the core accretion model. Low-mass planets cannot bind massive envelopes, while super-critical cores necessarily trigger runway gas accretion, leading to "forbidden" zones in the M-R plane. For a given mass, there is a considerable diversity of radii. We compare the synthetic M-R relation with the observed one, finding good agreement for a>0.1 AU. The synthetic radius distribution is characterized by a strong increase towards small R, and a second, lower local maximum at ~1 Jovian radius. The increase towards small radii reflects the increase of the mass function towards low M. The second local maximum is due to the fact that radii are nearly independent of mass for giant planets. A comparison of the synthetic radius distribution with Kepler data shows agreement for R>2 Earth radii, but divergence for smaller radii. We predict that in the next few years, Kepler should find the second, local maximum at ~1 Jovian radius.

Accepted to A&A. Minor revisions only relative to v1