A new and significant characterizing parameter (q(30)) which is easy to calculate is defined. This parameter can indicate in advance whether the PVT (pressure-volume-temperature) calculations performed on a crude oil using a predictive method will be in good agreement with experimental data. This parameter leads to a new classification of crude oils. When q(30) is small, the crude oil is considered as "classical" and all of its PVT properties can be accurately predicted. When q(30) is large, the crude oil is classified as "critical-like"; i.e., the prediction of some PVT properties may lead to significant deviations when compared to experimental values. In the latter case, a tuning process is proposed which in all cases improves the agreement between experimental and calculated data. Moreover, variance analysis theory is applied to estimate the experimental errors during a differential vaporization.