This study determined the validity of marginal ditching and staining as criteria for the diagnosis of secondary caries around amalgam restorations. One hundred and twenty-four Class I amalgam restorations on extracted human teeth were submitted to standardized clinical examinations. A calibrated examiner recorded the presence of ditching in the tooth/restoration interface, the occurrence of bluish-grey staining on the occlusal surface, and their relationship with the presence and severity of secondary carious lesions. Examinations were repeated after 4 days to calculate the intra-examiner reliability. Stereomicroscopic inspection of the serial tooth sections was used as the gold-standard. Six teeth were lost during the study. Of the 118 remaining teeth, 19 (16%) showed occlusal bluish-grey staining and 26 showed more than 0.20 mm wide marginal ditches. Of the latter, 14 presented narrow ditches (0.20-0.45 mm), while 12 presented wide ditches (0.45 mm). The evaluation criteria presented low sensitivity and positive predictive values. Positive and negative likelihood ratios showed that the criteria exhibited, indistinctly, the same odds to determine a positive or negative result for a diseased or healthy tooth. It was concluded that marginal ditching and staining were not valid criteria for the diagnosis of secondary caries around occlusal amalgam restorations.