On-site Raman analyses were performed at the Musée National de Céramique, Sèvres, France, on rare Iznik (former Nicaea) pottery produced from ∼1480 to ∼1620. A comparison was made with a series of shards. The town of production of these potteries was highly disputed in the 1980s and many questions still remain. The potential of glaze on-site analyses as a classification/dating tool was evaluated. The structure of the silicate glaze does not change with the sample (index of polymerization ∼0.5–0.8, indicating a lead silicate composition); characteristic SiO stretching mode doublet at ∼985 and 1030–1050 cm−1. By contrast, the corresponding signature of most of the ‘Kütahya’ wares peaks at ∼1070–1090 cm−1. The lowest index is measured for a brilliant overglazed red bole, according to a lower temperature of (post)firing. The different crystalline phases identified in the glaze are α-quartz, haematite, spinel, cassiterite, uvarovite garnet and zircon. White colour arises from α-quartz slip in most samples studied. Cassiterite (SnO2) opacifier is only present in some early blue-and-white ceramics (Master of the Knots and Baba Nakkas style, ∼1510–1530) and we do not have other evidence of its intentional use as an opacifier. Intentional addition of tin oxide is likely for colour lightening in some red, blue and in clear green boles. At least two types of red glazes and two types of Cr-containing green pigments are evidenced. Copyright © 2004 John Wiley & Sons, Ltd.