Two systems, and one of them is still everywhere
PINYIN is the standard romanisation of Mandarin and is what modern writing uses: Tieguanyin, Longjing, Dahongpao, Puer. WADE-GILES is the older system, and it survives in tea because the trade adopted it long before pinyin existed: Ti Kuan Yin, Lung Ching, Ta Hung Pao, Pu-erh. Older books, older vendors and a great deal of the second-hand market still use it. The differences are systematic rather than random — Wade-Giles k where pinyin has g, t where pinyin has d, ch where pinyin has zh or j — so once you have seen a few pairs the rest become guessable. Spacing is a third axis with no standard at all: Tieguanyin, Tie Guan Yin and Tie-Guan-Yin are all common and all correct.
The tea is often not named in Mandarin
TeaHQ's own reading of well-attested general knowledge, with no single source behind it.
This is the part that surprises people. Many famous teas come from regions where the local language is not Mandarin, and the name entered English through the local pronunciation. OOLONG is the clearest case: it is wu long in Mandarin, and the English word came through a southern Chinese pronunciation. BOHEA, CONGOU, SOUCHONG, PEKOE and HYSON are all seventeenth- and eighteenth-century English renderings of Min or Cantonese pronunciations, fossilised in the trade before anybody was standardising anything. "Pekoe" in the grading acronyms descends from a Min word for the white down on a bud. Cantonese, Min and Hakka pronunciations circulate alongside Mandarin ones for the same characters, which is why one tea can have four unrelated-looking English spellings that are all faithful to something.
Translation versus transliteration
A third axis: some sellers transliterate the sound and some translate the meaning, and a reader may not realise they are looking at the same tea. Tie Guan Yin / Iron Goddess of Mercy. Bai Hao Yin Zhen / Silver Needle. Da Hong Pao / Big Red Robe. Bi Luo Chun / Green Snail Spring. Long Jing / Dragon Well. Neither convention is wrong. Translations are more memorable in English and less searchable; transliterations are more searchable and opaque. Good vendors give both, and a vendor giving only the translated name is harder to cross-shop, which may or may not be deliberate.
Japanese and Korean are tidier, with two traps
TeaHQ's own reading of well-attested general knowledge, with no single source behind it.
Japanese romanisation is comparatively settled. The traps are macrons and their absence — hojicha and hōjicha, gyokuro and gyokurō conventions vary — and the older Kunrei system occasionally surfacing (sencha is stable; some terms are not). Korean has a bigger problem: the romanisation standard changed in 2000, so older writing uses the McCune-Reischauer system and newer writing the Revised system. Hadong is stable; other terms are not, and both spellings remain in circulation.
Searching for a tea in practice
Try the pinyin form without spaces, the pinyin form with spaces, the Wade-Giles form, and the English translation. Between them those four cover most of what is on sale. If a vendor uses only one convention consistently, note which — it tells you roughly when their catalogue vocabulary was formed, and older conventions frequently indicate an older business rather than an inferior one. And where a name is ambiguous in English, the Chinese characters resolve it instantly. A vendor who prints them is giving you the one unambiguous identifier available.