A flexible belt shearing and extrusion system for walnut shell crushing, kernel taking, and shell–kernel separation was used to study the effect of hulling and kernel taking from incompletely cracked walnuts. A one-factor test was conducted to explore the influences of upper–lower belt spacing (A), speed difference (B), and upper belt tension (C) on hulling. An orthogonal experiment was also designed to analyze the influences and determine the orthogonal design of the three factors. According to the experimental results, the influences of the three factors follow the trend of A>C>B for kernel exposing ratio I and II and C>A>B for kernel-breaking ratio and unreleased kernel ratio. The optimal hulling effect can be obtained with kernel exposing ratio I of 77%, kernel exposing ratio II of 16.7%, kernel breaking ratio of 4.3%, and unreleased kernel ratio of 2% (hulling ratio=98% and unbroken kernel ratio=93.7%) when A=14 mm, B=0.19 m/s, and C=231 N.