Recurrent drought in the Horn of Africa threatens water and food security. Integrated aquaculture-agriculture (IAA) systems are promoted for enhanced water productivity and resilience, but lack diagnostic tools for context-specific optimisation in water-scarce basins. This study aimed to develop and apply a diagnostic framework to evaluate IAA systems' performance and identify water management strategies for improving drought resilience in a semi-arid basin. A mixed-methods approach was employed, combining hydrological modelling, farm-level surveys, and water quality monitoring across multiple IAA sites. System performance was analysed using a stochastic frontier model: $Y_{it} = f(X_{it}; \beta) \exp(v_{it} - u_{it})$, where $u_{it}$ represents inefficiency. Robust standard errors were calculated to account for heteroscedasticity. The diagnostic framework revealed that integrated systems using sequential water reuse from ponds to crops increased total farm water productivity by an estimated 32% (95% CI: 24% to 40%) compared to non-integrated systems. A key theme was the critical role of small-scale water storage in buffering crop production during dry spells. IAA systems, when properly configured using a diagnostic approach, significantly enhance drought resilience through more productive water use and risk diversification at the farm level. Policy should support the adoption of diagnostic tools for IAA design and invest in farmer training on integrated water management. Extension services should prioritise water storage and reuse techniques. water productivity, stochastic frontier analysis, resilience, smallholder farmers, semi-arid, system integration This paper provides a novel diagnostic framework for quantifying water-use efficiency and resilience benefits in IAA systems, offering a replicable method for arid and semi-arid regions.