The conversion of coals to volatiles or liquids during pyrolysis and liquefaction is notoriously limited by the formation of retrograde products. Analysis of literature data for coals with grafted structures and for polymeric coal models demonstrates that the formation of volatile products from these materials does not correlate primarily with the weakness of the original bonding but correlates with the facility for retrogressive reaction. This analysis suggests further that simple recombination of resonance-stabilized radicals does not tend to yield true retrograde products, except in the case of aryloxy radicals. For pure hydrocarbon structural elements, radical addition to aromatic systems appears to be a key class of retrograde reactions, where the key factor is the kinetics of radical or H-atom loss from a cyclohexadienyl intermediate. We have used a mechanistic numerical model with a detailed set of radical reactions and thermochemically based kinetic parameters operating on a limited set of hydrocarbon structures to delineate important factors in mitigating retrograde processes. The modeling results show ( 1) how the "better" radical scavengers may reduce retrograde reaction at short reaction times but actually tend to increase it at longer times; ( 2) that the beneficial effects of H-2 pressure at short reaction times are not primarily due to lowering of harmful radical concentrations by scavenging, nor to the maintenance of donor content; ( 3) that the benefit is due to the small population of free H-atoms thus produced, which are very active in causing increased scission of strong bonds; and ( 4) that under some conditions retrograde products are actually generated faster with added H-2, but at longer reaction times and higher temperatures this temporary disadvantage of H-2 is overcome by increased hydrogenolysis of those earlier-produced retrograde products. Thus, not only the cleavage of critical bonds in the original coal structures but also the net prevention of retrogression may be due to the H-transfer-induced cleavage of strong bonds.