The aim of this study was to compare the accuracy of several common methods for the diagnosis of fissure caries. 63 human teeth without fillings and without any macroscopic carious cavitation but with different degrees of fissure discoloration and decalcification were selected from a pool. Dentists were asked to examine embedded teeth for fissure caries. The examination was done under standard conditions in a professional dental unit. The methods employed were: visual inspection (VI, n = 26 dentists), visual inspection with a magnifying glass (2x; VIM; n = 26), conventional bite-wing radiography (BW; n = 24), visual inspection combined with conventional bite-wing radiography (VI + BW; n = 10), and visual inspection combined with light pressure probing (VI + P; n = 23 dentists). In order to measure the reproducibility, VI and VI + BW were repeated. After the last inspection, the teeth were histologically prepared, serially sectioned perpendicular to the occlusal surface, and diagnosed for the presence of caries. The agreement between histological and clinical diagnosis was assessed. Specificities and sensitivities were: VI = 93 and 12%, VIM = 89 and 20%, BW = 83 and 45%, VI + BW = 87 and 49%, and VI + P = 93 and 14%, respectively. The following likelihood ratios were found: VI = 1.84, VIM = 1.86, BW = 2.6, VI + BW = 3.85, and VI + P = 2.05. It was concluded that the rather low sensitivity especially for VI and VI + P may lead to a significant number of teeth with dentinal caries being undetected.(ABSTRACT TRUNCATED AT 250 WORDS)