Abstract: Tactile and visual methods are not completely efficient in caries diagnosis. A visual method reinforced by a dye (Kuraray, Japan) was used to evaluate accuracy of the conventional methods. Two hundred and four teeth with cavities (class i and ii) already confirmed caries free by specialists in patients were chosen. The cavities dried and stained with a drop of the dye and were washed by water after 10 sec. Chi-square and z-tests were used for statistical analyses. The remained caries, especially in dentinoenamel junction (DEJ) stained dark pink color. Out of 102 class I cavities, 43 (42%) were stained in DEJ (10 in mesial, 14 in distal, 19 in buccal and 19 in lingual walls). Out of 102 class ii cavities, 49 (48%) were stained in DEJ (27 in gingival, 10 in lingual, 19 in buccal, 3 in mesial or distal (axial), 19 in buccal box and 8 in lingual box walls). The difference observed between class I and class ii cavities was not significant (p<0.05). In class I cavities no significant differences was observed among different parts involved (p<0.05). In class ii cavities the rate of caries was significantly higher in gingival wall (p<0.05). Out of 44 maxillary and 58 mandibular teeth of class I cavities 17 and 26 had remained caries, respectively. The difference was not significant (p<0.05). Out of 57 mandibular and 45 maxillary teeth of class ii cavities 34 and 15 had remained caries, respectively. The difference was significant (p<0.05). Totally the rate of caries was significantly higher in mandibular teeth (p<0.05). Accuracy of tactile and visual method was about 55%; therefore, more attention should be paid in diagnosis and removal of caries in DEJ. The use of caries detector dye was suggested.