Introduction: The existence of anatomical variations cause failures in endodontic treatments, so it is important to diagnose them. The objective was to determine the reproducibility and criterion validity of phosphor plate radiographs and sensor radiovisiography to identify anatomical variations detected by cone-beam computed tomography (CBCT) in lower premolars. Methods: 140 CBCT, radiographs and radiovisography premolar images were taken. Independent interpretations by two Endodontists were carried out to evaluate the Vertucci classification and teeth ramifications. Intra-observer and interobserver reproducibility were determined. Sensitivity, specificity and areas under the operator-receptor curve (AUC) were calculated using CBCT as the gold standard. Results: The intra-observer and interobserver reproducibility of the radiography was higher than the radiovisiography. For Vertucci Class I, radiography had higher sensitivity (94.7%), specificity (64.9%) and AUC (0.7952) than radiovisiography (89.3%, 62.2% and 0.7574, respectively), similarly in Class V (Radiography 69.2%, 93% and 0.8111; radiovisiography 50%, 84.2% and 0.6711, respectively). None of the techniques contributed to the diagnosis of Class III (AUC <0.5). Ramifications were infrequent (2.9%) and their detection was low (Sensitivity 25% for radiography and 0% for radiovisiography). Discussion: This is the first study thar evaluates the reproducibility and validity of these two radiographic techniques compared to CBCT for the detection of anatomical variations in teeth. Conclusions: Phosphor plate radiography showed higher reproducibility and validity for the diagnosis of Vertucci Class I and V, that were the most frequent observed premolar variations.