HIV attacks human immune cells, called T helper cells. Loss of these cells is gradual, often taking many years.
It was thought infected cells produced more HIV particles and that this caused the body to activate more T cells which in turn were infected and died.
Imperial College London modelling suggests that, if that was true, cells would die out in months not years.
The Imperial findings have been published in journal PLoS Medicine.
The imperial team used a mathematical model of the processes by which T cells are produced and eliminated.
Using this they showed that the current theory of an uncontrolled cycle of T cell activation, infection, HIV production and cell destruction - dubbed the "runaway" hypothesis - was flawed.
They concluded that it could not explain the very slow pace of depletion that occurs in HIV infection.
They showed that if the theory was correct, then T helper cell numbers would fall to very low levels over a number of months, not years.
Lack of certainty
Researcher Professor Jaroslav Stark said: "Scientists have never had a full understanding of the processes by which T helper cells are depleted in HIV, and therefore they've been unable to fully explain why HIV destroys the body's supply of these cells at such a slow rate.
"Our new interdisciplinary research has thrown serious doubt on one popular theory of how HIV affects these cells, and means that further studies are required to understand the mechanism behind HIV's distinctive slow process of cellular destruction."
The Imperial team thinks one possible explanation could be that the virus slowly adapts itself over the course of the infection.
But they stress that further analysis is needed to verify this alternative theory.
Professor Stark said: "If the specific process by which HIV depletes this kind of white blood cell can be identified, it could pave the way for potential new approaches to treatment."
Roger Pebody, a treatment advisor at HIV charity Terrence Higgins Trust, said: "HIV is an incredibly complex virus and research is ongoing to try and establish exactly how it works.
"We need more studies in this area before we can draw any clear conclusions."