The goal of our work was to design in a controlled manner open-shell supramolecular assemblies with dominant intermolecular ferromagnetic interactions. Along this line, the compounds of formula rad(6)M(CN)(6)I-3 . 2H(2)O with M = Cr, Fe, and Co have been synthesized; rad(+) stands for the 2-(3-N-methylpyridinium)-4,4,5,5-tetramethyl-4,5-dihydro-1H-imidazo-1-oxyl 3-N-oxide radical cation. The crystal structures have been solved. The three compounds are isomorphous. They crystallize in the trigonal space group P3 (a = 13.178(3) Angstrom, c = 10.763(6) Angstrom for M = Fe). The structure may be viewed as a two-dimensional honeycomb lattice, with three kinds of hexagons and three kinds of rad---rad magnetic interactions. In addition, for M = Cr and Fe, there is a fourth kind of magnetic interaction, occurring between the M(CN)(6)(3-) anion and the radical core. The magnetic properties of the three compounds have been investigated. They are characteristic of dominant ferromagnetic interactions to which weak antiferromagnetic interactions are superimposed. A quantitative interpretation of the magnetic data has been developed. In particular, for the chromium derivative, three out of four interactions have been found to be ferromagnetic, including the rad---Cr(CN)(6) interaction. The spin topologies of our supramolecular assemblies have been discussed in the context of the spin frustration phenomenon. The potentialities of the supramolecular chemistry of open-shell units have been emphasized.