In many forms of occupational exposure, the recorded mean dose rates from external radiation are much lower than the limit of 50 mSv in a year. In order to see whether the observance of this dose limit appeared to be influencing the magnitude of doses received, for example by constraining the year-to-year variation of annual doses, records of successive annual doses were examined for individuals exposed at different mean annual dose rates. Successive annual doses from external radiation in occupationally exposed individuals are approximately log-normally distributed around the mean dose rate at which the individual is exposed. The geometric standard deviation of annual doses, which estimates their variation in proportion to the dose rate, commonly decreases with increasing (geometric) mean dose rate. A dose rate is reached, however-typically at between 20 and 30 mSvy-1 in the data examined-at which the observed year-to-year variability of doses, if continued, would cause a dose limit of 50 mSv to be exceeded with a frequency of once in 40 worker-years of exposure at that dose rate. The records used in this review, which cover about 11000 annual doses received by 880 individuals, offer some quantitative assessment of the effect of the measures normally used to decrease the likelihood that annual doses might exceed the dose limit. Although the data only describe a limitation of such higher doses, rather than analysing the constraints which achieve this, they are of significance in suggesting that such constraints may apply even during exposure at low dose rates.