Objective To explore early features that can predict colchicine resistance in familial Mediterranean fever (FMF) patients. Methods It included FMF cases who fulfilled the Yalcinkaya-Ozen criterion and were on colchicine for at least 6 months. Data were collected from medical files and interpreted with respect to clinical parameters including, the auto-inflammatory diseases activity index (AIDAI) and FMF severity score. FMF50 score assessed the treatment response. Laboratory findings and genetic analysis of MEFV mutations were evaluated according to the standard technique. Patients were classified into two groups according to their response to colchicine. Both groups were compared and significant variables entered into a logistic regression model to detect independent predictors. The diagnostic accuracy of these predictors was assessed using the receiver operating characteristic (ROC) curve. Results 120 FMF children were included. After the exclusion of sixteen-non-complaint patients (13.3%), colchicine responders were 66 (63.4%) (group I), and colchicine resistant (group II) were 38 (36.5%). The fever duration after colchicine, number of attacks before/after colchicine, skin rash/erysipelas-like erythema, myalgia/protracted febrile myalgia, AIDIA before/after treatment, FMF severity score, and the maximum colchicine dose were higher in group II. Furthermore, high CRP and neutropenia were frequent in group II. However, different MEFV mutations including, M694V were similar between the two groups. Eight variables were detected in the regression analysis model and independent predictors were utilized to generate a scoring model. Conclusion : This study constructed a prediction model for colchicine nonresponse based on clinical and laboratory profiles. This model will be valuable for the treatment decisions of FMF children.