Multi-cluster hydraulic fracturing in highly deviated wells is a key stimulation technique for enhancing the development efficiency of low-permeability oil and gas reservoirs. To address engineering challenges such as cross-layer fracture propagation difficulties and uneven multi-cluster fracture growth during stimulation across multi-layered reservoirs, a numerical model for simulating multi-cluster fracture propagation in highly deviated wells was developed by combining the finite element method with the cohesive zone method, and its accuracy was verified against KGD (Khristianovic-Geertsma-de Klerk) analytical solution. On this basis, a “stimulation efficiency index” was proposed as a quantitative indicator to evaluate fracturing performance, taking into account both fracture growth uniformity and cross-layer propagation. The evolution characteristics and controlling factors of multi-cluster fractures in highly deviated wells were systematically investigated. The results show that increasing the well deviation angle appropriately, reducing the number of perforations, increasing the pumping rate, and lowering the number of clusters are beneficial to improving fracturing performance. Under low-viscosity slickwater conditions, the effect of fluid viscosity on fracture propagation is negligible. The relative importance of the parameters affecting stimulation performance is ranked as follows: number of clusters > pumping rate > number of perforations > deviation angle > fluid viscosity. Considering the geological and engineering characteristics of the target block, a stimulation efficiency index ≥0.8 is recommended as a criterion for optimal fracturing. The optimal design parameters are a pumping rate of ≥12 m³/min for four-cluster fracturing and ≥18 m³/min for six-cluster fracturing. Field microseismic monitoring confirmed the rationality and applicability of the proposed parameters.