This in vitro study aimed to design a method for the objective evaluation of carious dentin using numerical values. This study also investigated the relationship between the color of carious dentin stained with a caries detector dye using this objective method and the rate of bacterial detection as detected by a polymerase chain reaction (PCR). In 15 molars with occlusal dentin caries and three extracted sound molars, dentin was removed in multiple steps with 300 microm removed each step. Before and after every removal, images of a color-matching sticker and carious surfaces stained with a caries detector dye were acquired simultaneously using a CCD camera and dentinal tissue samples were removed with a round bur. Next, corrected L*, a* and b* values of the carious surfaces (CIE 1976 L*a*b* color system) were calculated from the color changes of the stickers in the images. In addition, bacterial DNA in the dentinal tissue was detected by PCR. From evaluations of the receiver operating characteristic curves for the L*, a* and b* values, the L* value was determined to be a more useful parameter than a* or b* for detecting bacterial infection using the caries detector dye. The bacterial detection rates of carious dentin decreased as the L* values of carious dentin stained with the dye increased. When the L* values were more than 60, the dentin had no bacterial infection. This study clarified the relationship between the colors of lesions stained with a caries detector dye and the rates of bacterial detection.