The Maya are the only people of the pre-Columbian Americas who left a script recording their language in full, which ought to make their religion the best documented in the hemisphere. Most of the writing is gone.
Four books survive, all Postclassic or later, named for the cities where they ended up: Dresden, Madrid, Paris, and a fourth long dismissed as a forgery and now generally accepted, held in Mexico. They are almanacs and ritual calendars, concerned with prognostication, deities and astronomy. The rest was lost to climate, to conquest, and to deliberate destruction. Diego de Landa, later bishop in Yucatan, burned manuscripts and images at Mani in 1562 in a campaign that also used torture, and afterwards wrote the account of Maya custom and script on which scholars have relied ever since, including the flawed alphabet that eventually helped to crack the writing.
Inscriptions on stone, wood, shell and painted pottery number in the thousands and are contemporary with the events they describe, which makes them the most reliable source available. They are also narrow: they record what rulers wanted recorded. Decipherment came slowly. The calendrical portions were understood by the late nineteenth century, but the language itself was not read until the phonetic approach proposed by Yuri Knorozov in 1952 and the demonstration by Tatiana Proskouriakoff in 1960 that the texts recorded the lives of real people rather than pure astronomy. Readings are still revised, and popular books written before the 1980s often rest on interpretations since abandoned.
Colonial and later sources fill gaps at a cost. The Popol Vuh and the Books of Chilam Balam were written in Maya languages in European script after the conquest, by authors already living under Christianity. Ethnography among living Maya communities records practice that is genuinely old and also five centuries removed from the Classic period. Reading it back onto the monuments is a standing temptation and a standing source of error.