Majority of persisting 3×3 patch based local methods developed simple and less effective methodologies for creating the feature size. Their results are not up to mark in unconstrained conditions. Furthermore, these descriptor imposes uniform coordination for pixels’ comparison. All these factors degrade the performance of descriptor. This work introduces the novel local descriptor ZigZag Local Binary Pattern (ZZ-LBP) in different unconstrained conditions with having effective methodology as compared to the existing ones. In ZZ-LBP, two zigzag designs are created and then as per the zigzag orientation, the pixels are compared. The codes attained from both designs are fused to develop ZZ-LBP size. ZZ-LBP size is on higher side therefore PCA is used in compacting the size. SVMs is used for matching. Testing is conducted on ORL and GT datasets. ZZ-LBP emerges as efficient and effective descriptor in comparison with other methods. For comparison, five local methods LBP, VELBP, MBP, NI-LBP and LMBP are evaluated with ZZ-LBP and rest are literature-based. ZZ-LBP secures the highest accuracy of 98.12% and 90.00% on ORL and GT.