Background and study aims: Endosonography is considered a valuable technique in the evaluation of pancreatic cysts. The aim of the present study is to assess interobserver agreement, in three different observer groups, regarding EUS for characterization of pancreatic cysts. Patients and methods: Video sequences of 40 EUS procedures for pancreatic cysts were prepared. Three groups of observers had different levels of EUS experience: group 1 comprised four experts with extensive EUS experience, group 2 had four "semi-experts" with limited EUS experience, and group 3 (novices) comprised four nonexpert resident physicians without EUS experience. Features scored included septations, nodules, solid components, and pancreatic duct communication. A presumptive diagnosis had to be specified. The intraclass correlation coefficient (ICC) was used, with agreement classed as excel-lent (> 0.80), good (0.61-0.80), moderate (0.41-0.60), fair (0.20-0.40), and poor (< 0.20). Results: Agreement regarding nodules was good among experts (ICC 0.65) and fair in the semi-expert and novice groups (ICC 0.32 and 0.37, respectively). For presence of solid components there was significantly higher agreement among experts (ICC 0.52) compared with the other two groups (semi-experts 0.09, and novices 0.03). Agreement regarding specific diagnosis was moderate in the expert group (0.43), poor among the semi-experts (0.09), and fair among the novices (0.30). Conclusions: Interobserver agreement among expert endosonographers was mostly moderate for characteristics of pancreatic cysts. However, interobserver agreement for experts was equal to or higher than that in the semi-expert and in the novice groups.