Polynucleotide phosphorylase was used as a probe to study the conformation of tRNA in solution. This enzyme degrades polyribonucleotide chains in the presence of orthophosphate into nucleoside diphosphate (phosphorolysis reaction) in a sequential and non-dissociating fashion (“processive” mechanism). All natural and biosynthetic polyribonucleotides, except tRNA, are completely degraded by the enzyme. Only a fraction of tRNA molecules can be phosphorolyzed at low temperatures: thus, at a given temperature some tRNA molecules are completely degraded from the 3′OH end, while the remaining molecules are entirely resistant, and their 3′OH terminal adenosine 5′-phosphate stays intact. This phenomenon has been extensively investigated and the results presented in this article lead to the conclusion that the reaction stops before complete phosphorolysis, not because the enzyme has become inactive, but because the limiting factor resides in the substrates themselves, i.e. tRNA. This property of tRNA is not a result of the formation of aggregates resistant to the enzyme; neither is it due to preferential phosphorolysis of certain species of tRNA, since the same phenomenon is observed with several purified specific tRNAs, as well as with bulk tRNA isolated from Escherichia coli, yeast, or rat liver. The results suggest the existence of two classes of tRNA conformations defined by their ability (S class) or inability (R class) to be phosphorolyzed. These conformations are not identical to the native and denatured forms, as defined by the ability or inability of tRNA to be charged with amino acid by aminoacyl synthetases. The effects of temperature, of mono-or divalent ions, and of denaturing agents were studied. Purine has been found to markedly enhance the percentage of phosphorolysis without any concommittant effect on the enzyme. Pre-heating of tRNA at a temperature higher than 90°C, in the presence of divalent ion can modify the percentage distribution of the two types of molecules. The conformational change of each molecule is temperature-dependent. The apparent non-interconvertibility of the two classes of conformations at low temperature is an unusual dynamic behavior which cannot be explained by a “frozen” equilibrium. It can, however, be accounted for in a multiple equilibrium model with a slow rate of interconversion involving large entropic changes.