Comparison of BCF models based on log P

Seven linear and nonlinear BCF models based on log P have been compared in order to estimate their accuracy, predictive power, and domain of application. This comparison was performed from a data set constituted of 436 experimental BCF values recorded for 227 chemicals. We showed that for chemicals with log P<6, the different models yielded equivalent results. At the opposite, for highly hydrophobic chemicals (log P>6), the bilinear model log BCF = 0.91 log P − 1.975 log (6.8 10−7 P + 1) − 0.786 (n = 154; R = 0.95; S = 0.347; F = 463.5) was superior to the other studied models.