Note to the seventh reprint
In this printing, a number of misprints have been corrected; I thank those readers who are kind enough to keep reporting them. I have also once again taken the opportunity of bringing the contents up to date in a few places, specifically new results on the supernova Hubble diagram (Fig. 5.4) and CMB anisotropies (Fig. 18.2).
Given the pace of cosmological research, I am surprised, but pleased, to see that the basic framework described in the original text survives without the need for revolutionary change. Nevertheless, some very significant developments have occurred since the first printing. Here is a personal list of recent highlights:
(1) Results on atmospheric neutrinos show that the $\mu$ neutrino oscillates, probably to a $\tau$ neutrino. If so, the $\tau$ neutrino mass is $\sim 0.06$~eV, and hot dark matter is unimportant (hep-ex/9912007). The SNO experiment has proved that oscillations are also the explanation for the solar neutrino problem (see e.g. hep-ph/0204314).
(2) The supernova Hubble diagram now argues very strongly for vacuum energy, and an accelerating expansion (see the new Fig. 5.4, and e.g. astro-ph/0701510). For a flat universe, the vacuum equation of state is within about 10% of $w=-1$.
(3) Small-scale CMB data show a clear set of acoustic peaks in the power spectrum (see the new figure 18.2 and the wonderful WMAP results in astro-ph/0603449). The combination of large-scale structure data (e.g. astro-ph/0501174) and the CMB requires a flat CDM model with $\Omega_m=0.24\pm0.02$ and $h=0.73\pm0.02$. This combined analysis shows exciting evidence for a tilted spectrum: $n=0.948\pm0.015$, or $n$ closer to unity but with significant tensor anisotropies. The simplest $V=m^2\phi^2$ inflationary model fits very well. The CMB has also been shown to be linearly polarized, at the expected level (astro-ph/0209478). In particular, this effect in the WMAP data gives a detection of scattering from reionization.
(4) Given that the $\Lambda$CDM model works so well on large scales, it is essential to understand smaller-scale galaxy correlations in detail. This is now possible using the `halo model' (e.g. astro-ph/0206508), which explains scale-dependent bias in terms of correlated pairs within a single halo, and between different haloes.
(5) Nevertheless, worries about the basic CDM paradigm persist, and are most severe on small scales, where the structure of galaxy-scale dark matter haloes has difficulty in matching observations (astro-ph/0312194). The nature of dark matter continues to be one of the greatest uncertainties in cosmology.
(6) Quasars have now been detected out to $z=6.42$, where the intergalactic medium is much less strongly ionized than at slightly lower redshift (astro-ph/0512082). We may be seeing close to the era of general reionization. Galaxies are now seen to significantly greater redshifts, on the verge of $z=7$, using the Lyman-break dropout technique and searches for Lyman-$\alpha$ emission (astro-ph/0411117 and astro-ph/0609393).
(7) The debate on initial conditions continues. Models with extra dimensions (the `brane world') have suggested alternatives to inflation (hep-th/0111030). A small cosmological constant presents a fine-tuning problem, and `quintessence' models attempt to solve this by using scalar-field dynamics (astro-ph/9901388; astro-ph/0304277). There is an increasingly good case, however, that the small non-zero vacuum density may need to be explained anthropically (astro-ph/0210358). This links nicely with progress in string theory, which predicts a `landscape' of many possible incarnations of low-energy physics (hep-th/0302219).