We present a determination of the optical luminosity function of quasi-stel
lar objects (QSOs) and its cosmological evolution with redshift for a sampl
e of over 6000 QSOs identified primarily from the first observations of the
2dF QSO Redshift Survey (2QZ), For QSOs with -26 < M-B < -23 and 0.35 < z
< 2.3, we find that pure luminosity evolution (PLE) models provide an accep
table fit to the observed redshift dependence of the luminosity function (L
F). The LF is best fitted by a two-power-law function of the form Phi(L-B)
proportional to [(L-B/L-B*)(alpha) + (L-B/L-B*)(beta)](-1). Exponential lum
inosity evolution models, both as a function of look-back time, L-B*(z) = L
-B*(0)e(k1 tau), and as a general second-order polynomial, L-B*(z) proporti
onal to 10(k1z+k2z2), were found to provide acceptable fits to the data set
comprising the 2QZ and the Large Bright Quasar Survey. Exponential evoluti
on with look-back time is preferred for q(0) = 0.05, while the polynomial e
volution model is preferred for q(0) = 0.5, The shape and evolution of the
LF at low redshifts (z < 0.5) and/or high luminosities, not currently well
sampled by the 2QZ survey, may show departures from pure luminosity evoluti
on, but the results presented here show that, over a significant range of r
edshift, PLE is a good description of QSO evolution.