The Balakot Formation of the Himalayan foreland basin in Pakistan was originally described as a >8 km thick clastic red bed succession within which are stratigraphically intercalated four gray marl bands containing fossils dated at 55--50 Ma. On this basis, and the reported conformable contact with the underlying Paleocene Patala Formation, the Balakot Formation red beds were interpreted as tidal facies dated at 55--50 Ma, by >20 Myr the oldest foreland basin sediments eroded from the metamorphic orogen. However, our new detailed structural mapping shows the Balakot Formation to be in tectonic contact with the underlying Patala Formation, and the marl bands to be structurally intercalated with the red beds. Hence neither the nature of the Balakot Formation lower contact nor the intercalated marl bands can be used to date the red beds. Our Ar/Ar dating of 257 individual detrital white micas from the Balakot Formation red bed sandstones shows that the red bed succession must be younger than 37 Ma, and we thus conclude that the first exposed foreland basin continental sediments eroded from the metamorphic mountain belt were deposited after 37 Ma, at least 15 Myr later than previously believed. These new structural, stratigraphic, and isotopic data from the Balakot Formation provide new constraints to the early tectonic evolution of the mountain belt, result in reassessment of the facies and provenance of the Balakot Formation, and force reconsideration of models of orogenesis, basin evolution, and the timing and diachroneity of India-Asia collision that are based on the original documentation of the Balakot Formation.