The C $IV$λλ1498,1501 broad emission line is visible in optical spectra to redshifts exceeding $z$ ∼ 5. C $IV$ has long been known to exhibit significant displacements to the blue and these ‘blueshifts’ almost certainly signal the presence of strong outflows. As a consequence, single-epoch virial black hole (BH) mass estimates derived from C $IV$ velocity widths are known to be systematically biased compared to masses from the hydrogen Balmer lines. Using a large sample of 230 high-luminosity ($LBol$ = 10$\mathrm{<mrow\; data-mjx-texclass="ORD">45.5</mrow>}$–10$\mathrm{<mrow\; data-mjx-texclass="ORD">48</mrow>}$ erg s$\mathrm{<mrow\; data-mjx-texclass="ORD">-1</mrow>}$), redshift 1.5 < $z$ < 4.0 quasars with both C $IV$ and Balmer line spectra, we have quantified the bias in C $IV$ BH masses as a function of the C $IV$ blueshift. C $IV$ BH masses are shown to be a factor of 5 larger than the corresponding Balmer-line masses at C $IV$ blueshifts of 3000 km s$\mathrm{<mrow\; data-mjx-texclass="ORD">-1</mrow>}$ and are overestimated by almost an order of magnitude at the most extreme blueshifts, ≳5000 km s$\mathrm{<mrow\; data-mjx-texclass="ORD">-1</mrow>}$. Using the monotonically increasing relationship between the C $IV$ blueshift and the mass ratio BH(C $IV$)/BH(H$\alpha$), we derive an empirical correction to all C $IV$ BH masses. The scatter between the corrected C $IV$ masses and the Balmer masses is 0.24 dex at low C $IV$ blueshifts (∼0 km s$\mathrm{<mrow\; data-mjx-texclass="ORD">-1</mrow>}$) and just 0.10 dex at high blueshifts (∼3000 km s$\mathrm{<mrow\; data-mjx-texclass="ORD">-1</mrow>}$), compared to 0.40 dex before the correction. The correction depends only on the C $IV$ line properties – i.e. full width at half-maximum and blueshift – and can therefore be applied to all quasars where C $IV$ emission line properties have been measured, enabling the derivation of unbiased virial BH-mass estimates for the majority of high-luminosity, high-redshift, spectroscopically confirmed quasars in the literature.