If quasars are distributed cosmologically according to their red-shifts, then it is expected that their distribution on the sky will be very close to random, because it is in that case an average over an enormous range in distance. Most studies find no significant clustering. It has recently been suggested, however, that there may be a highly significant excess of close pairs of quasars with different redshifts1, although this has been disputed2. The new ROE/ESO quasar survey3 provides a very large and deep sample over a connected area of sky, ideal for a more sensitive search for clustering. The result is shown in fig. 1. The observed distributions appear to be random, on all scales from less than an arcmin to tens of degrees.