Subjects made paired comparison judgements about the relative probabilities of pairs of nine hypothetical probabilities statements. Afterwards, the same subjects made subjective probability estimates for the statements. Then the estimates were used to generate pseudo paired comparison judgements, i.e., what the subjects would have said if their estimated subjective probabilities actually determined their paired comparison judgements. Subjective probability scales were derived from both kinds of data using Thurstone's Law of Comparative Judgement. Both derived scales closely corresponded to the estimates and to each other. It was concluded that both the paired comparison judgements and the estimates reflected subjects' underlying subjective probabilities for the statements but that estimates are a more efficient measure than are judgements.