Given the extensive use in periodontal diagnosis of the clinical sign of bleeding on probing, the reliability of the technique in relation to gingival histological features and the need for a better understanding of its significance was investigated. Bleeding (B) and non-bleeding (NB) buccal gingival tissues, were measured through manual probing during gingivectomy procedures. The histological evidence of inflammation was used as the gold standard. The established histological descriptive patterns showed statistically significant differences between B and NB sites (chi2 = 20.842, p < 0.05). The reliability scores were found to be of high sensitivity (90.9%) and negative predictive value (89.5%) as well as of fair specificity (77.3%) and positive predictive value (80%). The accuracy was found to be 84.1%. The histological quantitative analysis showed no statistically significant difference for any HGI (Histological Gingival Index). B sites presented HGI = 1 in 72.8%, close to the NB sites, which showed 63.4% for the same histological score. Furthermore, Pearson's analysis showed a low correlation between the studied clinical signs and the histological quantitative analysis (r = 0.265, p > 0.05). We conclude that the clinical parameter of bleeding on probing is a reliable tool for diagnosing gingival inflammation, whereas the clinical sign of non-bleeding on probing is not a safe criteria for evaluating gingival health.