For AIDS cases in Scotland the date of HIV antibody positive diagnosis (HIV test date) is recorded on the AIDS case reports. Data are also available on the total numbers of individuals with HIV positive test reports in Scottish laboratories in each year since testing became available. These data are incorporated into a model of the HIV epidemic described in terms of testing rates by calendar year. The HIV infection curve is modelled as a step function, and the testing rates are allowed to differ between steps. Results are presented for intravenous drug users (IDUs) and for homosexual/bisexual men. The estimated rates of testing for the IDUs (estimates from 19 per cent to 44 per cent per year) are considerably higher than those for homosexual/bisexual men (estimates of 10 per cent to 17 per cent per year). Data on the year of testing for AIDS diagnoses gave relatively little improvement in estimates of the HIV infection curve. However, when this information is combined with data on the total number of HIV positive diagnoses per year, there is a dramatic improvement in the estimate of the HIV infection curve. This is particularly marked for infections in the most recent period and for the estimates of cumulative infections up to the present. However, these improvements are gained at the cost of assumptions of similar testing rates applying to all sections of the HIV infected populations, which will be difficult to check in practice. This suggests that these methods should not be used in isolation but in combination with other evidence about the spread of HIV infection in a population.