The purpose of this in vitro study was to calibrate occlusal restorative decision-making based on visual inspection (VI), and to compare it with visual inspection with magnifying (x1.25) lenses (VIM), and fibre-optic transillumination (VIF). Sixty extracted human third molars mounted in plaster were examined using VI by ten dentists three times during the calibration process. After wash-out periods, the teeth were re-examined by the same dentists using VI, VIM and VIF. The inter-examiner reproducibility for VI, expressed as Kappa (kappa) statistics (kappa +/- SD), was 0.46 +/- 0.10 before and 0.59 +/- 0.11 after the calibration process and remained at the same level after the wash-out period. Intra-examiner reproducibility was substantially higher for VI (kappa -0.68 +/- 0.15) before and after the wash-out period. There were no significant changes in restorative treatment decisions based on VIM and VIF when compared to VI. The reproducibility between the methods was good for VI vs. VIM (kappa = 0.64) and moderate for VI vs. VIF (kappa = 0.56). In contrast, the inter-examiner reproducibilities expressed as Kappa were considerably lower for VIM (kappa = 0.46 +/- 0.17) and VIF (K = 0.42 +/- 0.19). It was concluded that the inter-examiner reproducibility can be improved with a calibration process based on a simple feedback method and that inter-examiner variation, especially without calibration, may be as important a factor for restorative treatment decisions as the diagnostic method itself.