ABSTRACTThe Miocene‐Pliocene Siwalik Group records changing fluvial environments in the Himalayan foreland basin. The Nagri and Dhok Pathan Formations of this Group in the eastern Potwar Plateau, northern Pakistan, comprise relatively thick (tens of metres) sandstone bodies and mudstones that contain thinner sandstone bodies (metres thick) and palaeosols. Thick sandstone bodies extend for kilometres normal to palaeoflow, and are composed of large‐scale stratasets (storeys) stacked laterally and vertically adjacent to each other. Sandstone bodies represent single or superimposed braided‐channel belts, and large‐scale stratasets represent channel bars and fills. Channel belts had widths of km, bankfull discharges on the order of 103 cumecs and braiding parameter up to about 3. Individual channel segments had bankfull widths, maximum depths, and slopes on the order of 102 m, 101 m and 10−4 respectively, and sinuosities around 1‐1. These rivers are comparable to many of those flowing over the megafans of the modern Indo‐Gangetic basin, and a similar depositional setting is likely. Thin sandstone bodies within mudstone sequences extend laterally for on the order of 102 m and have lobe, wedge, sheet and channel‐form geometries: they represent crevasse splays, levees and floodplain channels. Mudstones are relatively bioturbated/disrupted and represent mainly floodbasin and lacustrine deposition. Mudstones and sandstones are extremely disrupted in places, showing evidence of prolonged pedogenesis. These ‘mature’ palaeosols are m thick and extend laterally for km. Lateral and vertical variations in the nature of their horizons apparently depend mainly on deposition rate.The 500 m‐thick Nagri Formation has a greater proportion and thicker sandstone bodies than the overlying 700 m‐thick Dhok Pathan Formation. The thick sandstone bodies and their large‐scale stratasets thicken and coarsen through the Nagri Formation, then thin and fine at the base of the Dhok Pathan Formation. Compacted deposition rates increase with sandstone proportion (0‐53 mm/year for Nagri, 0‐24 mm/year for Dhok Pathan), and palaeosols are not as well developed where deposition rates are high. Within both formations there are 100 m‐scale variations (representing on the order of 105 years) in the proportion and thickness of thick sandstone bodies, and tens‐of‐m‐scale alternations of thick sandstone bodies and mudstone‐sandstone strata that represent on the order of 104 years. Formation‐scale stratal variations extend across the Potwar Plateau for at least 100 km, although they may be diachronous: however, 100‐m and smaller scale variations can only be traced laterally for up to tens of km.Alluvial architecture models indicate that increases in the proportion and thickness of thick sandstone bodies can be explained by increasing channel‐belt sizes (mainly), average deposition rate and avulsion frequency on a megafan comparable in size to modern examples. 100‐m‐scale variations in thick sandstone‐body proportion and thickness could result from ‘regional’ shifts in the position of major channels, possibly associated with ‘fan lobes’on a single megafan or with separate megafans. However, such variations could also be related to local changes in subsidence rate or changes in sediment supply to the megafan system.Formation‐scale and 100‐m‐scale stratal variations are probably associated with interelated changes in tectonic uplift, sediment supply and basin subsidence. Increased rates of hinterland uplift, sediment supply and basin subsidence, recorded by the Nagri Formation, may have resulted in diversion of a relatively large river to the area. Alternatively, changing river sizes and sediment supply rates may be related to climate changes affecting the hinterland (possibly linked to tectonic uplift). Climate during deposition of the Siwalik Group was monsoonal. Although the deposits contain no direct evidence for climate change, independent evidence indicates global cooling throughout the Miocene, and the possibility of glacial periods (e.g. around 10‐8 Ma, corresponding to base of Nagri Formation). If the higher Himalayas were periodically glaciated, a mechanism would exist for varying sediment supply to megafans on time scales of 104‐105 years. Although eustatic sea‐level changes are related to global climatic change, they are not directly related to Siwalik stratigraphic changes, because the shoreline was many 100 km away during the Miocene.