The purpose of this study was to evaluate different methods of eliciting gingival bleeding as indicators of gingival inflammation in the experimental gingivitis model. Following a period of stringent oral hygiene, 103 dental students were scored for plaque and gingival bleeding assessed by 4 methods. From this group, 41 volunteers were randomly allocated to 2 treatment groups. Dental students with clean teeth and healthy gingivae were asked to abolish all mechanical tooth cleaning in the lower jaw for a period of 3-weeks. During the 21-day experimental period, chlorhexidine (Peridex) or a placebo mouthrinse was applied to the lower jaw. Subjects brushed the upper jaw with a standard toothpaste. In principal, 2 different methods were employed to provoke bleeding: (1) at the marginal gingival tissue by running a probe along the soft tissue wall at the orifice of the pocket, and (2) by probing to the "bottom" of the pocket. Variations in the methods were based on angulation (AngBI, ParBI) of the probe in relation to the tooth surface and to the probing force (PPBI.25N, PPBI.75N). 1 randomly selected quadrant in the lower jaw was scored using the AngBI. The opposing quadrant was scored with a randomly-allocated bleeding index, either ParBI, PPBI.25N or PPBI.75N. The results of this study confirm earlier findings that the angulation of the probe determines the number of sites with bleeding observed. It also indicates that bleeding as elicited by probing to the bottom of the pocket is a poor indicator of early gingivitis. It is recommended that gingivitis should be assessed by probing the marginal gingiva.