The extent of the groin lymph node (LN) dissection for melanoma is still being debated, particularly in the case of micrometastasis (sentinel lymph node positive). We tested the predictive values of the criteria for pelvic dissection currently suggested by national guidelines (number of positive inguinal LN, Cloquet's LN status, and preoperative computed tomographic scan) and the inguinal lymph node ratio (LNR, the ratio between metastatic and excised LNs) to identify patients with pelvic metastasis. We analyzed the predictive values of the above-mentioned criteria in 157 patients who underwent an ilioinguinal dissection, with a focus on their negative predictive values (NPV), which might help identify low-risk patients who might safely avoid pelvic dissection, pelvic dissection reduction, and error rate. Forty-four (26.7%) patients had pelvic LN metastasis. In patients with micrometastasis (17.3% had pelvic LN metastasis), LNR less than 0.1 and Cloquet's LN status achieved clinically relevant NPV (95.7 and 95.5%, respectively) and pelvic dissection reduction (38.4 and 84.6%, respectively), whereas the error rate was 1.7 and 3.0%, respectively. Lower NPVs were observed for number of positive inguinal LNs (88.6%) and computed tomographic scan (78.4%). Accuracy was enhanced when these criteria were considered in multivariable models. In patients with macrometastasis (36.8% had pelvic LN metastasis), LNR and current selection criteria achieved low NPVs and a high error rate. Avoiding pelvic dissection may be safe in sentinel lymph node-positive patients with LNR less than 0.1. The prediction of pelvic metastasis seems to be less accurate for patients with clinically positive LNs.