Objective To examine temporal trends in the incidence of primary thyroid cancers diagnosed in 0–49 year olds in parts of Great Britain (GB) during the period 1976–2005. We specifically aimed to analyse age, period and cohort effects.

Design Population-based descriptive analysis of cancer registry data.

Setting Parts of Great Britain.

Participants Case data on thyroid cancer were obtained from four regional cancer registries in GB (i. Northern and Yorkshire, ii. North West, iii. Wales and iv. Scotland).

Main outcome measures Age-standardised incidence rates (ASRs) and 95% CIs were calculated. Negative binomial regression was used to examine the effects of age, sex, drift (linear trend), non-linear period and non-linear cohort.

Results The study analysed 4327 cases of thyroid cancer aged 0–49 years at diagnosis. For males, the overall ASR was 3.9 per million persons per year (95% CI 3.6 to to 4.1). For females, the overall ASR was 12.5 per million persons per year (95% CI 12.0 to to 12.9). The best fitting negative binomial regression model included age (p<0.001), sex (p<0.001) and drift (p<0.001). Non-linear period (p=0.42) and non-linear cohort (p=0.71) were not statistically significant. For males aged 0–14 years, the ASR increased from 0.2 per million persons per year in 1976–1986 to 0.6 per million persons per year in 1997–2005. For males aged 15–29 years and 30–49 years the ASRs increased from 1.9 to 3.2 and from 7.3 to 12.6 per million persons per year, respectively. For females aged 0–14 years, the ASR increased from 0.3 to 0.5 per million persons per year. For females aged 15–29 years and 30–49 years the ASRs increased from 7.0 to 12.3 and from 21.2 to 40.0 per million persons per year, respectively.

Conclusions There has been a linear increase in the incidence of thyroid cancer, which has led to a doubling of the number of cases diagnosed over a 20 year time span. The reasons for this increase are not well understood, but it is consistent with findings from other countries.