SUMMARY1 The changes in abundance ofAsterionella formosain the South Basin of Windermere between 1946 and 1990 are described and analysed. The average seasonal cycle for the 45‐year period shows an overwintering population of about 10 cell ml−1which increases with an exponential rate of 0.09 logeday−1to an annual maximum of 4000 cell ml−1by about Day 124. There is then a rapid decline at an exponential rate of loss of 0.29 logeday−1to values which typically are less than 0.01 cell ml−1in mid‐summer. By about Day 240 a second period of rapid increase occurs with an exponential rate of increase of 0.18 loge. day−1to a plateau of about 7 cell ml−1in late autumn and early winter.2 This average pattern is subject to considerable year‐to‐year variation. The timing and extent of the increase in the autumn was particularly variable. The rate of increase in the spring was strongly positively correlated, and that in the autumn strongly negatively correlated, with the day at which the exponential phase started. Rates for these two phases of increase were not statistically different when expressed in terms of time from mid‐summer, which reinforces earlier conclusions that light availability is the main factor governing the rate of spring increase and suggests that this is also the case for the autumn increase.3 Eight descriptors of seasonal development showed statistically significant changes over the 45 years. Early winter populations declined from 27 to 4 cell ml−1, and linked to this the day at which cell concentrations exceeded 50 cell ml−1occurred later by 24 days from Day 54 in 1946 to 78 in 1990. The lower early winter population appears to be linked to a lower end of year population as this decreased between 1946 and 1968 from 46 to 2 cell ml−1, and then increased slightly to 7 cell ml−1in 1990. The start of the spring exponential increase occurred on Day 57 in 1946 and started earlier by 19 days in 1968 but then occurred later, at Day 76, in 1990. The duration of the spring increase got shorter by 23 days, from 67 days in 1946 to 44 days in 1990. The maximum rate of increase rose from 0.065 logeday−1in 1946 to 0.112 logeday−1in 1990. The annual maximum declined from 9863 cell ml−1in 1946 to 2278 cell ml−1in 1968 and then increased to 6159 cell ml−1in 1990. The annual geometric mean decreased from 61 cell ml−1in 1946 to 5 cell ml−1in 1968 and remained nearly constant subsequently.4 In many cases, the precise underlying causes of these changes were not apparent. However, the increase with time of rate of increase in the spring appeared to be linked to a later start and hence growth under higher light. There was no signific