We study predictions in the concordance Lambda cold dark matter cosmology for the abundance and clustering of high-redshift galaxies and for the properties of their descendants. We focus on three high-redshift populations: Lyman break galaxies (LBGs) at z similar to 3, optically selected star-forming galaxies at z similar to 2 (BXs) and distant red galaxies ( DRGs) at z similar to 2. We select galaxies from mock catalogues based on the Millennium Simulation using the observational colour and apparent magnitude criteria. With plausible dust assumptions, our galaxy formation model can simultaneously reproduce the abundances, redshift distributions and clustering of all three observed populations. The star formation rates (SFRs) of model LBGs and BXs are lower than those quoted for the real samples, reflecting differing initial mass functions and scatter in model dust properties. About 85 per cent of model galaxies selected as DRGs are star forming, with SFRs in the range 1 to similar to 100 M(circle dot) yr(-1). Model LBGs, BXs and DRGs together account for less than half of all star formation over the range 1.5 < z < 3.2; many massive, star-forming galaxies are predicted to be too heavily obscured to appear in these populations. Model BXs have metallicities which agree roughly with observation, but model LBGs are only slightly more metal poor, in disagreement with recent observational results. The model galaxies are predominantly disc dominated. Stellar masses for LBGs and BXs are similar to 10(9.9) M(circle dot), and for DRGs are similar to 10(10.7) M(circle dot). Only about 30 per cent of model galaxies with M(*)>10(11)M(circle dot) are classified as LBGs or BXs at the relevant redshifts, while 65 per cent are classified as DRGs. Almost all model LBGs and BXs are the central galaxies of their dark haloes, but fewer than half of the haloes of any given mass have an LBG or BX central galaxy. Half of all LBG descendants at z = 2 would be identified as BXs, but very few as DRGs. Clustering increases with decreasing redshift for descendants of all three populations, becoming stronger than that of L* galaxies by z = 0, when many have become satellite galaxies and their typical stellar mass has increased by a factor of 10 for LBGs and BXs and by a factor of 3 for DRGs. This growth is dominated by star formation until z similar to 1 and thereafter by mergers. Merging is predicted to be more important for LBG and DRG descendants than for BX descendants. Most LBGs and DRGs end up as red ellipticals, while BXs have a more varied fate. Over 70 per cent of local galaxies with M(*) > 10(11)M(circle dot) are predicted to have at least one LBG/BX/DRG progenitor.
