The construct of alexithymia has been postulated as a predisposing factor in psychosomatic illness. The alexithymia construct has achieved wide currency in psychosomatic research and theorizing despite its doubtful psychometric foundations. Also, the question of between-group variability in alexithymia has not been addressed. In this study we proposed and tested a Rorschach measure of alexithymia on four groups of psychosomatic patients (back pain, gastrointestinal, dermatology, migraine headache). It was hypothesized that psychosomatic groups would be more alexithymic than non patients. To examine the question of between-group variability in alexithymia, it was hypothesized that back pain patients would be more alexithymic than other psychosomatic groups. Both hypotheses were supported. Additionally, exploratory comparisons between psychosomatic groups revealed a number of differences between the groups in basic personality processes. The findings show promise for the use of the Rorschach test as an alexithymia measure. Based on the current study, heterogeneous grouping of psychosomatic patients in research designs and treatment programs appear to be a highly questionable procedure.