Recent studies from Europe suggest a continuing increase in thyroid cancer, but it is unclear whether this trend also applies to the United States. The current study examined the long-term trend of thyroid cancer in Connecticut. Our results show that the overall age-adjusted incidence rate of thyroid cancer has been increasing in Connecticut, from 1.30/100,000 in 1935-1939 to 5.78/100,000 in 1990-1992 in females, and from 0.30/100,000 in 1935-1939 to 2.77/100,000 in 1990-1992 in males. The increase mainly comes from papillary carcinoma of the thyroid. The birth cohort analyses indicate that the increase in thyroid cancer occurred among cohorts born between 1915 and 1945, which experienced an increase of 31.4% every 5 years in males and 17.3% in females over the period 1960-1979. For those born since the 1945 cohort, the incidence has been decreasing, at rates of 9.3% and 8.3% every 5 years over the period 1975-1992 in males and females, respectively. Age-period-cohort modeling results also suggest a strong birth cohort effect on the observed time trend in both sexes, which closely follows the introduction of radiation treatment of benign childhood conditions in the head and neck between 1920 and the 1950s in the United States. Our results are consistent with the suggested radiation hypothesis, indicating that radiation treatment of benign childhood conditions in the head and neck is largely responsible for the observed increase of thyroid cancer in Connecticut.