La velocidad con que los bifenilos policlorados (PCBs) son eliminados en humanos, representada por sus períodos de vida media (t1/2), es de gran relevancia ambiental. En este trabajo, a partir de datos reportados en la literatura, se elabora una base de datos de esta variable tóxico-cinética de PCBs en humanos. 14 estructuras de PCBs son optimizadas energéticamente, haciendo previamente un barrido del ángulo diedro entre sus anillos fenilícos , utilizando AM1 para encontrar la conformación más estable. Los PCBs mono-orto sustituidos presentan un ángulo diedro aproximado de 120° y los poli-orto sustituidos de 90°. A partir de las estructuras optimizadas se calculan diversos descriptores moleculares, empleando el método B3LYP/6-31G**. Los descriptores son asociados con los t1/2 mediante análisis de regresión lineal múltiple. Este análisis revela que el t1/2 de PCBs es función de la sumatoria de cargas de átomos de cloro (ΣQCl) y el valor total de cargas de carbonos en posición meta (ΣQC-meta). El coeficiente de regresión es de 0.84 y PThe elimination rate of the polychlorinated biphenyls (PCBs) in humans is of great environmental importance, as represented by their half-lives (t1/2). Based on the literature, a database of this toxickinetic variable for PCBs in humans is elaborated. 14 PCB structures with known half-lives are optimized, making a previous scan of the dihedral angle between the two phenyl rings, using AM1 to find the most stable conformation. Substituted mono-orto PCBs present an approximate dihedral angle of 120°, and the substituted poly-orto of 90°. Several descriptors are calculated from optimized structures using B3LYP/6-31G**. The molecular descriptors are associated with the t1/2 values by multiple lineal regression analysis. This reveals that the t1/2 of PCBs is a function of the sum of atomic charges on the chlorine atoms (ΣQCl) and the sum of atomic charges for the carbon atoms in meta position (ΣQC-meta). The regression coefficient value of the prediction model is 0.84, and the Pvalue is <0.01 (range: 1.2-24 years). A discriminant analysis allows to verify the molecular characteristics that describe best the t1/2 of PCBs. The discrimination function carried out for two groups, finds that the descriptor ΣQC-meta is the most important to classify the PCBs according to their t1/2. Taking in account that the metabolism of PCBs depends on the substitution of the carbon atoms in meta-para positions, since this begins with the formation of an epoxy-PCB in these atoms of the biphenyls, the electronic descriptor ΣQC-meta can explain this local chemical environment and the metabolic susceptibility of the congener to degradation by cytochrome P450 isozymes (CYPs).