The mechanisms causing AIDS and subsequently death in human immunodeficiency virus type 1 infection are not yet fully understood. Nonetheless, correlates of accelerated progression to disease based on immunological and virological markers have been identified. The best correlate identified to date is the baseline virus load or the so-called viral set point. By focusing on a virus load measurement from a restricted time range, however, we ignore valuable information contained in the long-term profile of the virus load. Here, we investigate the relationship between virus load and survival with the aid of a statistical model. The model takes into consideration the virus load at every stage of the disease. In particular, we aim to determine the effect of peaks of virus load on disease progression. We fit our model to unique sequential viral load data of 12 simian immunodeficiency virus mac251-infected rhesus macaques which contain frequent measurements throughout the entire course of the infection until the development of simian AIDS. Our model enables us to predict the survival times of the animals more accurately than an equivalent model which considers the viral set point only. Furthermore, we find that peaks of the virus load contribute less to disease progression than phases of low virus load with the same amount of viral turnover. Our analysis implies that the total viral turnover is not the best correlate of survival. As a consequence, the direct cytopathic effects of virus replication may, by themselves, have less of an impact on disease progression than previously thought.