We have used VLT FLAMES data to constrain the uncertain physics of rotational mixing in stellar evolution models. We have simulated a population of single stars and find two groups of observed stars that cannot be explained: (1) a group of fast rotating stars which do not show evidence for rotational mixing and (2) a group of slow rotators with strong N enrichment. Binary effects and fossil magnetic fields may be considered to explain those two groups. We suggest that the element boron could be used to distinguish between rotational mixing and the binary scenario. Our single star population simulations quantify the expected amount of boron in fast and slow rotators and allow a comparison with measured nitrogen and boron abundances in B-stars.