This study numerically investigates the chaotic mixing induced by chevron patterns and examines its influence on enhanced reverse osmosis (RO) filtration. The primary design variables include the dimensionless groove depth (dp*) and the dimensionless overlapped span (S*) of the patterns. The Reynolds number (Re) varies from 100 to 1,000, within laminar flow conditions. We analyze the mixing characteristics using Poincaré sections and quantify mixing progress through the intensity of segregation (Id). For each Re value, a specific combination of dp* and S* induces chaotic mixing. Filtration performance is evaluated based on the permeate flux, total flow rate through the membrane, and Sherwood number. Our study reveals that both chaotic advection and intensified lateral flows near the patterns are essential for enhanced RO filtration. Patterns that induce globally chaotic mixing do not always correlate with optimal filtration performance, highlighting that the degree of chaotic mixing is not directly related to concentration polarization (CP) mitigation or filtration enhancement. Rather, it is the intense lateral flow near the patterns that acts as the decisive factor in achieving superior filtration performance, particularly at higher S* values (S*≥0.75) and with deeper grooves (dp*≥0.2). In terms of energy efficiency, patterned membranes demonstrate superior performance compared to their flat counterparts.