There now exist several data sets with the depth necessary to determine the power spectrum of galaxy clustering on scales greater-than-or-similar-to 100 Mpc. Results from redshift surveys and angular clustering all agree on the form of the function, which shows a break towards greater uniformity at very large wavelengths, consistent with scale-invariant primordial fluctuations. The amplitude of these large-wavelength fluctuations implies that the Cold Dark Matter model should use a higher normalization; an increase by a factor 3 is allowed by present microwave-background limits. Such a revised amplitude predicts too much small-scale power in linear theory, but this discrepency is on a scale where astrophysical effects are expected to alter the shape of the power spectrum. Nevertheless, if a better match between data and linear theory is desired, models containing global texture perform extremely well.