The effect of a shear flow on the critical behavior of a binary fluid (cyclohexane-aniline) has been investigated by light-scattering techniques. Turbidity and scattered-light-intensity measurements have been made, and the influence of all the independent parameters of the phenomenon were considered, i.e., temperature $T$, shear rate $S$, transfer wave vector $|\stackrel{\ensuremath{\rightarrow}}{\mathrm{q}}|$, and its projection ${q}_{X}$ along the flow direction $X$. The data support the Onuki-Kawasaki (OK) description, namely, (i) an effect exists only in the region $S\ensuremath{\tau}g1$, with $\ensuremath{\tau}$ the lifetime of concentration fluctuation. (ii) The critical temperature ${T}_{c}$ is lowered by the shear, so that ${T}_{c}={T}_{c}(S)={T}_{c}(0)\ensuremath{-}{T}_{0}{S}^{{\ensuremath{\sigma}}_{0}}$, with ${\ensuremath{\sigma}}_{0}=\frac{1}{3\ensuremath{\nu}}$ and $\ensuremath{\nu}=0.630$ the standard exponent of the correlation length in fluids. We find experimentally ${T}_{0}=(1.8\ifmmode\pm\else\textpm\fi{}0.2)\ifmmode\times\else\texttimes\fi{}{10}^{\ensuremath{-}4}$ and ${\ensuremath{\sigma}}_{0}=0.53\ifmmode\pm\else\textpm\fi{}0.03$, to be compared with the OK values 1.28 \ifmmode\times\else\texttimes\fi{} ${10}^{\ensuremath{-}4}$ and 0.529. (iii) The susceptibility ${\ensuremath{\chi}}_{\stackrel{\ensuremath{\rightarrow}}{\mathrm{q}}}$ shows an anisotropy versus $X$, and follows a mean-field behavior versus temperature. The OK dependence $\ensuremath{\chi}_{\stackrel{\ensuremath{\rightarrow}}{q}}^{}{}_{}{}^{\ensuremath{-}1}\mathrm{v}\ensuremath{\simeq}A{S}^{{\ensuremath{\sigma}}_{1}}{[T\ensuremath{-}{T}_{c}(S)]}^{\ensuremath{\gamma}}+B{S}^{{\ensuremath{\sigma}}_{2}}{q}_{X}^{\ensuremath{\omega}}+{q}^{2}$ fits the data well, with ${\ensuremath{\sigma}}_{1}\ensuremath{\simeq}0.14\ifmmode\pm\else\textpm\fi{}0.05 [\mathrm{OK}:\frac{(2\ensuremath{\nu}\ensuremath{-}1)}{3\ensuremath{\nu}}=0.137]$, $\ensuremath{\gamma}\ensuremath{\simeq}1.00\ifmmode\pm\else\textpm\fi{}0.06 (\mathrm{OK}:\ensuremath{\gamma}=1)$, $\ensuremath{\omega}\ensuremath{\simeq}0.40\ifmmode\pm\else\textpm\fi{}0.05 (\mathrm{OK}:\frac{2}{5})$, except for ${\ensuremath{\sigma}}_{2}$, whose experimental value ${\ensuremath{\sigma}}_{2}\ensuremath{\simeq}1\ifmmode\pm\else\textpm\fi{}0.15$ is about twice the OK value 8/15. Discrepancies of about a factor 2 are also found for the amplitudes $A$ and $B$.