The new potentially octadentate ligand, 1-(carboxymethyl)-4,7-bis[(6-carboxypyridin-2-yl)methyl]-1,4,7-triazacyclononane (H(3)bpatcn), in which two picolinate arms and one acetate arm are connected to the 1,4,7-triazacyclonane core, has been prepared. Potentiometric studies show an increased stability of the Gd(III) complex of H(3)bpatcn (logK(GdL)=15.8(2)) with respect to the Gd(III) complex of the analogous ligand 1,4,7-triazacyclononane-N,N',N''-triacetic acid (H(3)nota) (logK(GdL)=13.7), associated with an increased selectivity of H(3)bpatcn for gadolinium over calcium. The H(3)bpatcn ligand sensitises the terbium ion very efficiently, leading to a long-lived and highly luminescent terbium complex (quantum yield=43 %), in spite of the presence of a coordinated water molecule. (1)H proton NMR studies indicate that the metal ion is rigidly encapsulated by the three arms of the octadentate ligand H(3)bpatcn and that the macrocycle framework remains bound (through the five nitrogen and the three oxygen atoms) even at high temperature. A new theoretical method for interpreting the water proton relaxivity is presented. It is based on recent progresses in the description of the electronic spin relaxation and on an auxiliary probe solute. It replaces the Solomon, Bloembergen and Morgan (SBM) framework, which is questionable at low field, while avoiding resorting to simulations and/or sophisticated theories with additional unknown zero-field splitting (ZFS) parameters. The inclusion of two picolinate groups on a triazacyclononane framework affords the mono-aquo gadolinium complex [Gd(bpatcn)(H(2)O)] with favourable electron-relaxation properties (tau(eff)(S0)=125 ps). The optimisation of the electronic relaxation by ligand design is especially important to achieve high relaxivity in the new generation macromolecular complexes with long rotational correlation times.