Maize production is central to food security and rural livelihoods in Tanzania, yet remains highly vulnerable to climate variability. Existing national-scale studies often overlook temporal dynamics or employ models with limited interpretability, restricting their practical utility. To address these limitations, we integrated a rigorously validated ARIMAX(3,2,3) time series model with Sobol’ global sensitivity analysis using annual data from 1961–2022. The ARIMAX model demonstrated strong predictive performance ( , MAPE ), indicating negative effects of relative humidity on maize production, positive contributions from precipitation and soil moisture, and weak direct temperature influence. Crucially, Sobol’ analysis revealed temperature as the dominant driver of production variability, accounting for of independent variance and with interactions, while precipitation primarily affects maize production through interactions. This integrated approach uncovers that short-term production responds to moisture conditions, but long-term variability is governed by thermal stress and complex climate interactions. By quantifying both direct and interactive weather effects, our study enhances methodological transparency and provides actionable guidance for improving maize resilience through heat management, soil moisture conservation, and humidity control. These results support climate-resilient agricultural strategies vital for sustaining food security in Tanzania and similar regions.