Background: This study was undertaken to retrospectively analyze the accuracy of different methods in differentiating binary atresia from neonatal jaundice. Methods: A search was made in MEDLINE, and the Web of Science for relevant original articles published in English; methodological quality of the included studies was also assessed. Two reviewers extracted data independently. Studies were pooled, summary receiver operating characteristics curve and diagnostic odds ratio (DOR) with corresponding confidence intervals were calculated. Results: For diagnosis of binary atresia, ultrasonography (US), hepatic scintigraphy (BIBS), and magnetic resonance cholangiography (MRCP) had a pooled sensitivity of 74.9% (range: 70.4%-79.1%), 93.4% (range: 90.3%-95.7%) and 89.7% (range: 84.8%-93.4%), a specificity of 93.4% (range: 91.4%-95.1%), 69.2% (range: 65.1%-73.1%) and 64.7% (range: 58.0%-71.0%), a positive likelihood ratio of 12.16 (range: 6.41-23.08), 3.01 (range: 2.15-4.20) and 3.10 (range: 1.59-6.06), a negative likelihood ratio of 0.23 (range: 0.13-0.38), 0.13 (range: 0.06-0.25) and 0.16 (range: 0.06-0.44), DOR of 72.56 (range: 27.34-192.58), 29.88 (range: 12.82-69.64) and 32.48 (range: 8.22-128.29), with an area under the curve of 0.96, 0.91, and 0.92, and Q value of 0.90, 0.85, and 0.85, respectively. Conclusions: US, HBS and MRCP can be very useful for the diagnostic work-up of neonatal cholestasis. To improve the sensitivity and specificity, several additional measures can be used.