Turkey comes out cheaper on the ground, with one caveat: the evidence is one resort. The Post Office Family Holiday Report 2026 (July 2026 prices, in pounds) prices the same basket of ten family items in 22 resorts: coffee, beer, cola, wine, water, suncream, insect repellent, children's snacks, a two-course lunch for two and a three-course family dinner. Marmaris, the only Turkish resort in it, came to £158.33. The Greek islands it measured came to £181.38 on Kos, £186.10 on Rhodes, £204.34 on Corfu and £221.57 on Crete. That makes Marmaris about 13% cheaper than Kos, 15% than Rhodes, 23% than Corfu and 29% than Crete.
Most of the gap is the family dinner: £83 in Marmaris against £92 on Rhodes and £106 on Corfu. Coffee is dearer in Marmaris (£3.41 against £2.27 on Rhodes), and wine and lunch are close, so it isn't cheaper on everything. The caveat is that this is one resort against four islands, and the basket leaves out the hotel and flights. The direction is steady, though: in the 2025 edition Marmaris was also the cheapest of 16 destinations, and Crete, Kos and Rhodes came 7th, 8th and 9th. We have no survey figures for Antalya, Bodrum or Fethiye, so we can't say the whole country is cheaper. The numbers also move fast: the report found Marmaris up 22.7% in a year in pounds, even as the lira fell by almost 12%.
Turkey's currency is the lira, and inflation was 29.73% in September 2026, down from 49.38% in September 2024, so prices in lira rise fast and any price list ages quickly. Many of the big sights are priced in euros for foreigners, so sightseeing costs about the same as in Greece: Hagia Sophia is 25 EUR, Ephesus about 40 EUR, the Acropolis about 30 EUR. Greece uses the euro, so what you read is what you pay.
The package is still the number that decides most trips. Price the same week at the same star rating in both countries on the dates you'd travel, with flights, and compare the total. If it's all-inclusive, add up what you'd spend on drinks and meals in a Greek resort to see what you're really comparing.