This study compared the intra-examiner and inter-examiner error of 2 constant force probes to the reading of a conventional manual probe. 3 examiners made repeated examinations of attachment level using a modified Florida probe and a manual North Carolina probe (read to 1 mm or 0.5 mm); relative attachment level measurements were made using a Florida disk probe. One probe was used in each quadrant in 8 subjects with moderate to advanced periodontitis. Error was calculated as the mean of the absolute value of the difference between each examination, and the correlation between values at each examination calculated. Statistically-significant differences between probe type, examiners, and sites were detected using a repeated measures ANOVA accounting for the nesting within subjects. There was a significant difference in error by probe type (modified Florida probe 0.62±0.03 mm, r=0.86; Florida stent probe 0.55±0.05 mm, r= 0.82; manual probe to 1 mm 0.39±0.02 mm, r=0.88; manual probe to 0.5 mm 0.40±0.02 mm, r=0.89;(p<0.001). Significant differences were observed by examiners (p<0.01). These data indicate that both manual and controlled-force probes can provide measurement within less than 1 mm of error; however, individual calibration of examiners remains important in the reduction of error. © Munksgaard, 1997.