Seroconversion illness is known to be associated with more rapid HIV disease progression. However, symptoms are often subjective and prone to recall bias. We describe symptoms reported as seroconversion illness and examine the relationship between illness, HIV test interval (time between antibody-negative and anibody-positive test dates) and the effect of both on time to AIDS from seroconversion. We used a Cox model, adjusting for age, sex, exposure group and year of estimated seroconversion. Of 1820 individuals, information on seroconversion illness was available for 1244 of whom 423 (34%) reported symptomatic seroconversion. Persons with a short test interval ([les ]2 months) were significantly more likely to report an illness than people with a longer interval (OR 6·76, 95% CI 4·75–9·62). Time to AIDS was significantly faster (P=0·01) in those with a short test interval. The HIV test interval is a useful replacement for information on seroconversion illness in studies of HIV disease progression.