Forty-nine individual carbazoles were identified in a fluid-catalytic-cracking light cycle oil (LCO) using gas chromatography (GC) with nitrogen chemiluminescence detection (GC-NCD). Their retention times were correlated with molecular descriptors generated from their molecular structures. The best seven-parameter multilinear-regression models showed good fit with a standard deviation of s = 0.58 min for the retention times. The descriptors involved in the models reflect the geometrical, topological, and electronic properties of the molecules, related to the interactions between the solute and the stationary phase. For forty-five out of the forty-nine carbazoles (carbazole, alkylcarbazoles, and hydrocarbazoles) of most interest in LCO processing, another seven-parameter multilinear-regression model was developed with standard deviation s = 0.37 min. This model further improved the model fit to the database of retention times. QSPR models of this quality can be used as an important source of information in the process of identification of unknown chromatographic peaks by matching their retention times with those of carbazoles of known molecular structures when the corresponding chemical standards are unavailable.