Defect detection is critical for ensuring the safety and reliability of ferromagnetic components, while accurate defect-depth evaluation is required to assess damage severity and support maintenance decisions. This study investigates the use of magnetic Barkhausen noise (MBN) signals and hysteresis loop, or BH curve measurements for evaluating defect depth in ferromagnetic materials. The theoretical framework is based on the magnetisation process and magnetic domain-wall motion. Under identical testing conditions, the presence of defects is expected to disturb the local magnetic response and reduce the effective magnetic domain activity, resulting in a weaker MBN signal. Simulations are first conducted to analyse BH curve variations caused by defects with different depths, providing guidance for subsequent experimental verification. Two types of defects are considered in the experiments: crack defects and welding defects. Different sensor probes are designed and applied according to the defect type and measurement requirements. The results show that variations in defect depth produce consistent trends in the simulated BH curves and experimentally measured MBN signals. These findings confirm that MBN measurements are sensitive to defect-depth variations and are consistent with the theoretical analysis. Furthermore, defect mapping based on MBN signal features provides an effective indication of defect distribution and depth, demonstrating the potential of the proposed magnetic method for non-destructive defect evaluation in ferromagnetic components.