All data in logBB database is based on in vivo measurements taken from rat studies. The whole data set includes 109 diverse organic compounds, which was divided into a training set of 72 compounds and two test sets of 37 compounds. The molecular geometries of all compounds were fully minimized using molecular mechanism. It should be noted that in refence  six compounds in the training set are duplicates to six compounds in the test set and are eliminated from the training set.
Download the databases:
training_set.sdf (the SDF flile includes 72 molecules)
test_set.sdf(the SDF file includes 37 molecules)
. Tingjun Hou, Xiaojie Xu, ADME Evaluation in Drug Discovery. 3. Modeling Blood-Brain Barrier Partitioning Using Simple Molecular Descriptors. Journal of Chemical Information and Computer Sciences, 2003, 43, 2137-2152. [html] [PDF]