"The molecular descriptor is the final result of a logic and mathematical procedure which transforms chemical information encoded within a symbolic representation of a molecule into a useful number or the result of some standardized experiment." (Handbook of Molecular Descriptors, R.Todeschini and V.Consonni, WileyVCH, 2000).
Molecular descriptors play a fundamental role in chemistry, pharmaceutical sciences, environmental protection policy, health research and quality control, they being obtained when molecules are transformed into a molecular representation allowing some mathematical treatment. Many molecular descriptors have been proposed derived from different theories and approaches with the aim of predicting biological and physicochemical properties of molecules [R.Todeschini and V.Consonni, Handbook of Molecular Descriptors, WileyVCH, Weinheim (GER), 2000].
The information content of a molecular descriptor depends on the kind of molecular representation that is used and on the defined algorithm for its calculation. There are simple molecular descriptors derived by counting some atomtypes or structural fragments in the molecule, other derived from algorithms applied to a topological representation (molecular graph) and usually called topological or 2Ddescriptors, and there are molecular descriptors derived from a geometrical representation that are called geometrical or 3Ddescriptors.
All the molecular descriptors must contain, to varying extents, chemical information, must satisfy some basic invariance properties and general requirements, and must be derived from wellestablished procedures which enable molecular descriptors to be calculated for any set of molecules. It is obvious – almost trivial  that a single descriptor or a small number of descriptors cannot wholly represent the molecular complexity or model all the physicochemical responses and biological interactions. As a consequence, although we must get used to living with approximate models (nothing is perfect!), we have to keep in mind that "approximate" is not a synonym of "useless".
The field of molecular descriptors is strongly interdisciplinary and involves a variety of different theories. For the definition of molecular descriptors, a knowledge of algebra, graph theory, information theory, computational chemistry, theories of organic reactivity and physical chemistry is usually required, although at different levels. For the use of the molecular descriptors, a knowledge of statistics, chemometrics, and the principles of the QSAR/QSPR approaches is necessary in addition to the specific knowledge of the problem.
The 29 logical molecular descriptors blocks (and their subblocks) calculated by Dragon are:
1.  Constitutional descriptors 
3.1.  Vertex degreebased indices 
3.2.  Distancebased indices 
4.2.  Selfreturning walk counts 
5.1.  KierHall molecular connectivity indices 
5.2.  Solvation connectivity indices 
5.3.  Randiclike connectivity indices 
6.2.  Indices of neighborhood symmetry 
7.  2D matrixbased descriptors 
7.2.  Topological distance matrix 
7.5.  Reciprocal squared distance matrix 
7.7.  Distance/detour matrix 
8.1.  BrotoMoreau autocorrelations 
8.2.  Centred BrotoMoreau autocorrelations 
8.3.  Moran autocorrelations 
8.4.  Geary autocorrelations 
8.5.  Topological charge autocorrelations 
10.  P_VSAlike descriptors 
10.4.  Van der Waals volume 
10.5.  Sanderson electronegativity 
10.7.  Ionization Potential 
12.  Edge adjacency indices 
12.2.  Connectivitylike indices 
13.  Geometrical descriptors 
13.3.  Delocalizationdegree indices 
14.  3D matrixbased descriptors 
14.1.  Geometrical distance matrix 
14.2.  Reciprocal squared geometrical distance matrix 
14.3.  Distance/distance matrix 
15.1.  TDB autocorrelations 
16.3.  Weighted by van der Waals volume 
16.4.  Weighted by Sanderson electronegativity 
16.5.  Weighted by polarizability 
16.6.  Weighted by ionization potential 
17.3.  Weighted by van der Waals volume 
17.4.  Weighted by Sanderson electronegativity 
17.5.  Weighted by polarizability 
17.6.  Weighted by ionization potential 
18.1.  Directional descriptors 
20.  Randic molecular profiles 
21.  Functional group counts 
22.  Atomcentred fragments 
23.  Atomtype Estate indices 
25.1.  Weighted topological atom pairs 
25.3.  Frequency Atom Pairs 
26.1.  Weighted geometrical atom pairs 
