Twitter is a medium where, when used adequately, users’ interests can be derived from what he follows. This characteristic can make it attractive for a source of personality derivation. We set out to test the hypothesis that, analogous to the Lexical hypothesis, which posits that word use should reveal personality, following behavior on social media should reveal personality aspects. We used a two-step approach, wherein the first stage, we selected accounts for whom it was possible to infer personality profiles to some extent using available literature on personality and interests. On these accounts, we trained a regression model and segmented the derived features using hierarchical cluster analysis. In the second stage, we obtained a small sample of users’ personalities via a questionnaire and tested whether the model from stage 1 correlated with the users from step 2. The the explained variance for the neurotic and neutral neuroticism groups indicated significant results (R2 = .131, p = .0205; R2 = .22, p = .0044). Confirming the hypothesis that following behavior should be correlated with one’s interests and that interests are correlated with the neuroticism personality dimension.