Background: When non-diabetic kidney disease (NDKD) is suspected, biopsy proven is used for definite diagnosis. However, there are not always easily available and may lead to cause complications. A clinical prediction score may help selecting appropriate patients for kidney biopsy. Objective: To develop a clinical prediction score for distinguishing any type of NDKD (NDKD alone or coexisting NDKD and diabetic nephropathy [DN]) and DN alone. Materials and Methods: A retrospective cohort study was conducted in type 2 diabetes mellitus (T2DM) patients with atypical features of DN, who had kidney biopsy at Thammasat University Hospital between 2011 and 2019. The present study divided patients into NDKD alone, coexisting NDKD and DN, and DN alone, confirmed by pathological diagnoses. The authors developed a clinical prediction score by weighing coefficients of predictors in a multivariable logistic model. Internal validation was performed with bootstrapping. Results: The present study included 81 patients of which 28 (34%) had NDKD alone, 15 (18%) had coexisting NDKD and DN, and 38 (41%) had DN alone. Primary membranous nephropathy, primary focal segmental glomerulosclerosis (FSGS), and secondary FSGS were prevalent in any NDKD. Absence of diabetic retinopathy (DR) showed a significant association with any NDKD (adjusted OR 3.72; 95% CI 1.28 to 10.8; p=0.02). The prediction score, AUROC of 0.75 (95% CI 0.63 to 0.86), had four predictors, duration of DM of less than 10 years, eGFR of more than 30 mL/minute/1.73 m², HbA1c of less than 8%, and absence of DR. Higher scores were associated with higher probability of NDKD. Conclusion: The present study clinical prediction score appears to be a useful tool to determine NDKD probability. T2DM patients with atypical presentation of DN with lower scores (0 to 2) may defer kidney biopsy. Keywords: Non-diabetic kidney disease; clinical prediction score; kidney biopsy; type 2 diabetes mellitus