Sediment dynamics and rates of landscape denudation in dryland regions are difficult to quantify due to long residence times of sediment on hillslopes and floodplains, often exceeding millions of years. Such extended durations complicate the interpretation of single cosmogenic nuclide analyses. This study applies paired Be-10 and Al-26 measurements in detrital samples to constrain basin-wide denudation rates and sediment fluxes within the semi-arid Upper Fortescue catchment, located in the Pilbara, northwestern Australia. Morphometric data, lithology and dual cosmogenic nuclide concentrations are integrated to quantify long-term basin-wide denudation rates and evaluate first-order controls on sediment production, transport and catchment evolution. A stepwise series of corrections were conducted for topographic shielding, lithology and apparent sediment burial. Corrected nuclide concentrations yield basin denudation rates in the Upper Fortescue catchment that range from 0.83 to 3.02 m/Ma. These values are comparable to channel bedrock rates (2.5 +/- 0.8 m/Ma; n = 4) and moderately higher than estimates from mesa summits and alluvial fan surfaces (0.8 +/- 0.6 m/Ma; n = 13) previously derived from Mn-53 measurements. Nine of eleven detrital samples exhibit Al-26/Be-10 ratios between 4.7 and 5.8, lower than the nominal production ratio of 6.75, indicating that sediments must have experienced a complex exposure history. Two conceptual scenarios are proposed to explain these ratios: (a) extended burial in colluvium or channel sediments lasting several hundred thousand years, and/or (b) complex exposure as a result of production rate attenuation at depth in slowly eroding (<5 m/Ma) bedrock, when sub-surface rock become exposed, probably through spalling of large meter-sized blocks from vertical cliff surfaces along escarpments and gorges, ubiquitous in the region. In scenario (b), most of our data are explained by materials being sourced from the average depth of 0.5 to 1 m. However, progressive downstream transport of channel sediments increases the likelihood of storage at shallow depth for a prolonged period in riverbanks, colluvium and floodplains. Field evidence combined with cosmogenic nuclide data supports a hybrid model involving both scenarios. These findings demonstrate the effectiveness of dual-nuclide analysis in quantifying long-term catchment-scale denudation and provide new insights into sediment source-to-sink processes in arid landscapes.