After the unexpected appearance of lethal symptoms on tomato plants infected with the PSTVd strain Intermediate Di, viroids were isolated and sequenced. It was found that a new strain, named RG 1, had been generated spontaneously in our greenhouse. In a different series of plant passages two new strains, named QF A and QF B, were detected which coexisted with the wild-type strain Di. Strains QF A and QF B showed intermediate symptoms when inoculated separately. In order to confirm the working hypothesis that the more pathogenic strain outcompetes the less pathogenic strain but strains of similar pathogenicity might coexist in the host, strains of different pathogenicity were mixed for inoculation in a ratio from 1:1 to 1:100 (more pathogenic:less pathogenic). The concentrations of the individual strains were determined 6 weeks postinfection with the method of nondenaturing polyacrylamide gel electrophoresis, and the working hypothesis was confirmed. The total concentrations of viroids in infected plants were very similar, irrespective of whether severe, intermediate, or mild strains or mixtures of different strains were present. The mutations in all new strains (3 in RG 1, 2 in QF A, 3 in QF B) were located in the so-called virulence-modulating region. The mutations of strain RG 1 influenced dramatically the thermodynamic stability of the native rod-like structure, as determined experimentally by temperature-gradient gel electrophoresis. Since during replication a multihairpin structure is generated transiently which is transformed afterwards into the rod-like structure, a lower thermodynamic stability of the rod-like structure leads to a higher accumulation of the transient structure. It is assumed that the transient structure, which is active in replication as shown earlier, is essential also in pathogenesis. This model explains the experimentally determined interdependence between pathogenicity and replicability of PSTVd strains.