We present a power spectrum analysis of the 10K catalogue from the 2dF QSO Redshift Survey . Although the Survey currently has a patchy angular selection function , we use the Virgo Consortium ’ s Hubble Volume simulation to demonstrate that we are able to make a useful first measurement of the power spectrum over a wide range of scales . We compare the redshift-space power spectra of QSOs to those measured for galaxies and Abell clusters at low redshift and find that they show similar shapes in their overlap range , 50-150h ^ { -1 } Mpc , with P _ { QSO } ( k ) \propto k ^ { -1.3 } . The amplitude of the QSO power spectrum at z \approx 1.4 is almost comparable to that of galaxies at the present day if \Omega _ { m } =0.3 and \Omega _ { \Lambda } =0.7 ( the \Lambda cosmology ) , and a factor of \approx 3 lower if \Omega _ { m } =1 ( the EdS cosmology ) is assumed . The amplitude of the QSO power spectrum is a factor of \approx 10 lower than that measured for Abell clusters at the present day . At larger scales , the QSO power spectra continue to rise robustly to \approx 400 h ^ { -1 } Mpc , implying more power at large scales than in the APM galaxy power spectrum measured by Baugh & Efstathiou . We split the QSO sample into two redshift bins and find little evolution in the amplitude of the power spectrum , consistent with the result for the QSO correlation function . In models with \Omega _ { m } \mathrel { \hbox { \hbox to 0.0 pt { \hbox { \lower 4.0 pt \hbox { $ \sim$ } } } % \hbox { $ > $ } } } 0.1 this represents evidence for a QSO-mass bias that evolves as a function of time . We compare the QSO power spectra to CDM models to obtain a constraint on the shape parameter , \Gamma . For two choices of cosmology ( \Omega _ { m } =1 , \Omega _ { \Lambda } =0 and \Omega _ { m } =0.3 , \Omega _ { \Lambda } =0.7 ) , we find the best fit model has \Gamma \approx 0.1 \pm 0.1 . In addition , we have shown that a power spectrum analysis of the Hubble Volume \Lambda CDM mock QSO catalogues with \Gamma = 0.17 as input , produces a result which is statistically consistent with the data . The analysis of the mock catalogues also indicates that the above results for \Gamma are unlikely to be dominated by systematic effects due to the current catalogue window . We conclude that the form of the QSO power spectrum shows large-scale power significantly in excess of the standard CDM prediction , similar to that seen in local galaxy surveys at intermediate scales .