A 160-m-long section measured in the lagoonal facies of the Middle Triassic Latemar platform (Dolomites, Italy) reveals a set of frequency components that we interpret as a strong Milankovitch signal. In this interpretation, all principal frequencies associated with the theoretical Middle Triassic precession index, P1 = 1/(21.7 ky), P2 = 1/(17.6 ky), and its modulations, E1 = 1/(400 ky), E2 = 1/(95 ky), and E3 = 1/(125 ky), were detected in a time–frequency evaluation of the cycles. A weak obliquity signal is also present in part of the section. Thus, the Latemar cycles appear to have recorded the clearest orbital forcing signal yet found in a carbonate platform. This astronomical calibration indicates that the section was deposited in ca. 3.1 My and therefore that the entire Latemar cyclic succession (~ 470 m) took at least 9 My to form. However, the calibration also leads to serious conflicts with other interpreted geological data: U/Pb radiometric ages of zircons collected from tuffites within the Latemar lagoon and in coeval basinal sediments point to a timescale that is five times shorter than this astronomically calibrated estimate; similar discrepancies arise when the average duration of Triassic ammonoid biozones or the sedimentation rates of coeval basinal series are considered. Nonetheless, all of the methods that have been used to estimate the time of formation of the Latemar platform continue to have shortcomings, and the contradictions among these different geological calibrations remain unresolved.