Background: Although the Toronto Alexithymia , a representative measure of alexithymia, was originally designed to measure three factors, recent factor analysis studies have consistently identified a two-factor structure, which merges DIF (Difficulty Identifying Feelings) and DDF (Difficulty Describing Feelings) factors into one and separates the EOT (Externally Oriented Thinking) factors. Methods: Meta-analysis of correlations among the three subfactors of TAS-20, as well as between the three subfactors of TAS-20 and the criterion variable (depression). Results: Correlation between DIF and DDF factors was consistently high, while correlation between EOT factor and the DIF and DDF factors combined was generally low. Correlation between criterion variable (depression) and the DIF and DDF factors combined was significant, while correlation between the criterion variable (depression) and the EOT factors was low or non-significant. The results were the same across five groups that are diagnostically disparate from each other (addiction, depression/anxiety, eating disorders, medical disease, healthy subjects). Conclusions: The findings of the meta-analysis indirectly support the recently raised alternative two-factor structure of TAS-20, suggesting that the EOT factor is heterogeneous from the other factors. We suggest separating the EOT subfactor scores from the total scores for proper utilization of TAS-20.