The objective of the study was to examine the association between fish and shellfish intake and diabetes in an island population, and the design of the study was Cross-sectional. Two independent population-based field surveys were conducted in Hvar Island of the eastern Adriatic coast of Croatia in May 2007 and May 2008, with a total of 1,379 adult participants. In multivariable logistic regression models, total fish intake was positively associated with diabetes prevalence in the total population (OR(Q4 vs. Q1) = 1.64; 95% CI = 1.01-2.66; p-trend = 0.09). Oily fish intake also exhibited a positive association with diabetes prevalence in the total population (OR(Q4 vs. Q1) = 2.22; 95% CI = 1.35-3.64; p-trend = 0.01) and in analyses stratified by body mass index, males and those with a high waist circumference. The study suggests an association between oily fish intake and diabetes in the population of the Hvar Island in Croatia. Longitudinal studies incorporating measures of persistent organic pollutants and local cooking practices are warranted to identify factors in fatty fish that may influence the development or persistence of diabetes.