In this paper we present a clustering analysis of QSOs over the redshift range z = 0.3 - 2.9 .
We use a sample of 10558 QSOs taken from preliminary data release catalogue of the 2dF QSO Redshift Survey ( 2QZ ) .
The two-point redshift-space correlation function of QSOs , \xi _ { Q } ( s ) , is shown to follow a power law on scales s \simeq 1 - 35 ~ { } h ^ { -1 } ~ { } { Mpc } .
Fitting a power law of the form \xi _ { Q } ( s ) = ( s / s _ { 0 } ) ^ { - \gamma } to the QSO clustering averaged over the redshift interval 0.3 < z \leq 2.9 we find s _ { 0 } = 3.99 ^ { +0.28 } _ { -0.34 } ~ { } h ^ { -1 } ~ { } { Mpc } and \gamma = 1.58 ^ { +0.10 } _ { -0.09 } for an Einstein-de Sitter cosmology .
The effect of a significant cosmological constant , \lambda _ { 0 } , is to increase the separation of QSOs , so that with \Omega _ { 0 } = 0.3 , \lambda _ { 0 } = 0.7 the power law extends to \simeq 60 ~ { } h ^ { -1 } ~ { } { Mpc } and the best fit is s _ { 0 } = 5.69 ^ { +0.42 } _ { -0.50 } ~ { } h ^ { -1 } ~ { } { Mpc } and \gamma = 1.56 ^ { +0.10 } _ { -0.09 } .
These values , measured at a mean redshift of \bar { z } = 1.49 , are comparable to the clustering of local optically selected galaxies .
We compare the clustering of 2QZ QSOs to generic CDM models with shape parameter \Gamma _ { eff } .
Standard CDM with \Gamma _ { eff } = 0.5 is ruled out in both Einstein-de Sitter and cosmological constant dominated cosmologies , where \Gamma _ { eff } \simeq 0.2 - 0.4 and \Gamma _ { eff } \simeq 0.1 - 0.2 respectively are the allowable ranges .

We measure the evolution of QSO clustering as a function of redshift .
For \Omega _ { 0 } = 1 and \lambda _ { 0 } = 0 there is no significant evolution in comoving coordinates over the redshift range of the 2QZ .
QSOs thus have similar clustering properties to local galaxies at all redshifts we sample .
In the case of \Omega _ { 0 } = 0.3 and \lambda _ { 0 } = 0.7 QSO clustering shows a marginal increase at high redshift , s _ { 0 } being a factor of \sim 1.4 higher at z \simeq 2.4 than at z \simeq 0.7 .
Although the clustering of QSOs is measured on large scales where linear theory should apply , the evolution of QSO clustering does not follow the linear theory predictions for growth via gravitational instability ( rejected at the > 99 per cent confidence level ) .
A redshift dependent bias is required to reconcile QSO clustering observations with theory .
A simple biasing model , in which QSOs have cosmologically long lifetimes ( or alternatively form in peaks above a constant threshold in the density field ) is acceptable in an \Omega _ { 0 } = 1 cosmology , but is only marginally acceptable if \Omega _ { 0 } = 0.3 and \lambda _ { 0 } = 0.7 .
Biasing models in which QSOs are assumed to form over a range in redshift , based on the Press-Schechter formalism , are consistent with QSO clustering evolution for a minimum halo mass of \sim 10 ^ { 12 } M _ { \odot } and \sim 10 ^ { 13 } M _ { \odot } in an Einstein-de Sitter and cosmological constant dominated universe , respectively .
However , until an accurate , physically motivated , model of QSO formation and evolution is developed , we should be cautious in interpreting the fits to these biasing models .